From 39b060ef225b62d41ec4cf9c615cdbbc4e715c11 Mon Sep 17 00:00:00 2001
From: Rene Brun <Rene.Brun@cern.ch>
Date: Fri, 17 Nov 2006 16:00:03 +0000
Subject: [PATCH] New version of TMVA fixing many coding conventions
violations. New version of the tmva test suite. To execute it run the script
TMVAnalysis.C
git-svn-id: http://root.cern.ch/svn/root/trunk@16805 27541ba8-7e3a-0410-8455-c3a389f83636
---
tmva/Module.mk | 1 +
tmva/inc/BinarySearchTree.h | 10 +-
tmva/inc/BinarySearchTreeNode.h | 6 +-
tmva/inc/BinaryTree.h | 6 +-
tmva/inc/CrossEntropy.h | 4 +-
tmva/inc/DataSet.h | 96 ++++----
tmva/inc/DecisionTree.h | 12 +-
tmva/inc/DecisionTreeNode.h | 6 +-
tmva/inc/Event.h | 6 +-
tmva/inc/Factory.h | 12 +-
tmva/inc/GeneticANN.h | 6 +-
tmva/inc/GeneticBase.h | 4 +-
tmva/inc/GeneticCuts.h | 6 +-
tmva/inc/GeneticGenes.h | 4 +-
tmva/inc/GeneticPopulation.h | 4 +-
tmva/inc/GeneticRange.h | 4 +-
tmva/inc/GiniIndex.h | 4 +-
tmva/inc/IMethod.h | 10 +-
tmva/inc/MethodBDT.h | 30 +--
tmva/inc/MethodBase.h | 60 ++---
tmva/inc/MethodBayesClassifier.h | 4 +-
tmva/inc/MethodCFMlpANN.h | 8 +-
tmva/inc/MethodCFMlpANN_Utils.h | 6 +-
tmva/inc/MethodCFMlpANN_def.h | 6 +-
tmva/inc/MethodCommittee.h | 10 +-
tmva/inc/MethodCuts.h | 36 +--
tmva/inc/MethodFisher.h | 14 +-
tmva/inc/MethodHMatrix.h | 10 +-
tmva/inc/MethodLikelihood.h | 10 +-
tmva/inc/MethodMLP.h | 34 +--
tmva/inc/MethodPDERS.h | 10 +-
tmva/inc/MethodRuleFit.h | 6 +-
tmva/inc/MethodSVM.h | 4 +-
tmva/inc/MethodTMlpANN.h | 6 +-
tmva/inc/MethodVariable.h | 6 +-
tmva/inc/Methods.h | 6 +-
tmva/inc/MisClassificationError.h | 4 +-
tmva/inc/MsgLogger.h | 46 ++--
tmva/inc/Node.h | 6 +-
tmva/inc/Option.h | 6 +-
tmva/inc/PDF.h | 10 +-
tmva/inc/Ranking.h | 8 +-
tmva/inc/Reader.h | 8 +-
tmva/inc/RootFinder.h | 6 +-
tmva/inc/Rule.h | 4 +-
tmva/inc/RuleEnsemble.h | 10 +-
tmva/inc/RuleFit.h | 4 +-
tmva/inc/RuleFitParams.h | 4 +-
tmva/inc/SdivSqrtSplusB.h | 4 +-
tmva/inc/SeparationBase.h | 4 +-
tmva/inc/SimulatedAnnealingBase.h | 6 +-
tmva/inc/SimulatedAnnealingCuts.h | 6 +-
tmva/inc/TActivationChooser.h | 22 +-
tmva/inc/TNeuron.h | 4 +-
tmva/inc/TNeuronInputChooser.h | 10 +-
tmva/inc/TSpline1.h | 6 +-
tmva/inc/TSpline2.h | 6 +-
tmva/inc/Timer.h | 6 +-
tmva/inc/Tools.h | 6 +-
tmva/inc/Types.h | 50 ++--
tmva/inc/VariableInfo.h | 26 +-
tmva/inc/Volume.h | 6 +-
tmva/src/BinarySearchTree.cxx | 10 +-
tmva/src/BinarySearchTreeNode.cxx | 6 +-
tmva/src/BinaryTree.cxx | 6 +-
tmva/src/CrossEntropy.cxx | 4 +-
tmva/src/DataSet.cxx | 208 +++++++++++-----
tmva/src/DecisionTree.cxx | 48 ++--
tmva/src/DecisionTreeNode.cxx | 8 +-
tmva/src/Event.cxx | 28 ++-
tmva/src/Factory.cxx | 70 +++---
tmva/src/GeneticANN.cxx | 8 +-
tmva/src/GeneticBase.cxx | 8 +-
tmva/src/GeneticCuts.cxx | 8 +-
tmva/src/GeneticGenes.cxx | 4 +-
tmva/src/GeneticPopulation.cxx | 4 +-
tmva/src/GeneticRange.cxx | 6 +-
tmva/src/GiniIndex.cxx | 4 +-
tmva/src/MethodANNBase.cxx | 30 ++-
tmva/src/MethodBDT.cxx | 104 +++++---
tmva/src/MethodBase.cxx | 63 ++++-
tmva/src/MethodBayesClassifier.cxx | 15 +-
tmva/src/MethodCFMlpANN.cxx | 16 +-
tmva/src/MethodCFMlpANN_Utils.cxx | 8 +-
tmva/src/MethodCommittee.cxx | 74 +++---
tmva/src/MethodCuts.cxx | 77 ++++--
tmva/src/MethodFisher.cxx | 29 ++-
tmva/src/MethodHMatrix.cxx | 43 +++-
tmva/src/MethodLikelihood.cxx | 31 ++-
tmva/src/MethodMLP.cxx | 39 ++-
tmva/src/MethodPDERS.cxx | 198 ++++++++-------
tmva/src/MethodRuleFit.cxx | 36 ++-
tmva/src/MethodSVM.cxx | 13 +-
tmva/src/MethodTMlpANN.cxx | 21 +-
tmva/src/MethodVariable.cxx | 10 +-
tmva/src/MisClassificationError.cxx | 4 +-
tmva/src/MsgLogger.cxx | 39 +--
tmva/src/Node.cxx | 6 +-
tmva/src/Option.cxx | 6 +-
tmva/src/PDF.cxx | 8 +-
tmva/src/Ranking.cxx | 29 ++-
tmva/src/Reader.cxx | 46 ++--
tmva/src/RootFinder.cxx | 6 +-
tmva/src/Rule.cxx | 77 +++---
tmva/src/RuleEnsemble.cxx | 88 +++++--
tmva/src/RuleFit.cxx | 40 ++-
tmva/src/RuleFitParams.cxx | 56 +++--
tmva/src/SdivSqrtSplusB.cxx | 4 +-
tmva/src/SeparationBase.cxx | 4 +-
tmva/src/SimulatedAnnealingBase.cxx | 34 +--
tmva/src/SimulatedAnnealingCuts.cxx | 15 +-
tmva/src/TNeuron.cxx | 4 +-
tmva/src/TSpline1.cxx | 6 +-
tmva/src/TSpline2.cxx | 6 +-
tmva/src/Timer.cxx | 6 +-
tmva/src/Tools.cxx | 25 +-
tmva/src/Types.cxx | 37 +--
tmva/src/VariableInfo.cxx | 10 +-
tmva/src/Volume.cxx | 8 +-
tmva/test/BDT.C | 189 +++++++++++++++
tmva/test/TMVAGui.C | 109 +++++++++
tmva/test/TMVAlogon.C | 57 +++++
tmva/test/TMVAnalysis.C | 363 +++++++++++++---------------
tmva/test/TMVAnalysis.py | 257 ++++++++++++++++++++
tmva/test/TMVApplication.C | 176 ++++++++++++++
tmva/test/annconvergencetest.C | 67 +++++
tmva/test/compareanapp.C | 155 ++++++++++++
tmva/test/correlations.C | 37 +--
tmva/test/correlationscatters.C | 160 ++++++++++++
tmva/test/efficiencies.C | 134 +++++-----
tmva/test/likelihoodrefs.C | 181 ++++++++++++++
tmva/test/line-small.png | Bin 0 -> 294 bytes
tmva/test/mutransform.C | 137 +++++++++++
tmva/test/mvas.C | 87 ++++---
tmva/test/network.C | 313 ++++++++++++++++++++++++
tmva/test/plotall.C | 23 ++
tmva/test/sigmoid-small.png | Bin 0 -> 363 bytes
tmva/test/tmvaglob.C | 267 ++++++++++++--------
tmva/test/variables.C | 298 ++++++++++++-----------
139 files changed, 4009 insertions(+), 1547 deletions(-)
create mode 100644 tmva/test/BDT.C
create mode 100644 tmva/test/TMVAGui.C
create mode 100644 tmva/test/TMVAlogon.C
create mode 100644 tmva/test/TMVAnalysis.py
create mode 100644 tmva/test/TMVApplication.C
create mode 100644 tmva/test/annconvergencetest.C
create mode 100644 tmva/test/compareanapp.C
create mode 100644 tmva/test/correlationscatters.C
create mode 100644 tmva/test/likelihoodrefs.C
create mode 100644 tmva/test/line-small.png
create mode 100644 tmva/test/mutransform.C
create mode 100644 tmva/test/network.C
create mode 100644 tmva/test/plotall.C
create mode 100644 tmva/test/sigmoid-small.png
diff --git a/tmva/Module.mk b/tmva/Module.mk
index 14b3fd7f8de..825dcc07ed6 100644
--- a/tmva/Module.mk
+++ b/tmva/Module.mk
@@ -3,6 +3,7 @@
#
# Author: Fons Rademakers, 20/6/2005
+
MODDIR := tmva
MODDIRS := $(MODDIR)/src
MODDIRI := $(MODDIR)/inc
diff --git a/tmva/inc/BinarySearchTree.h b/tmva/inc/BinarySearchTree.h
index bcd42733054..91f1495ec66 100644
--- a/tmva/inc/BinarySearchTree.h
+++ b/tmva/inc/BinarySearchTree.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: BinarySearchTree.h,v 1.17 2006/11/02 08:03:18 helgevoss Exp $
+// @(#)root/tmva $Id: BinarySearchTree.h,v 1.19 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -104,8 +104,8 @@ namespace TMVA {
// create the search tree from the events in a TTree
// using the variables specified in "theVars"
Int_t Fill( const DataSet& ds, TTree* theTree, Int_t theType = -1,
- Types::PreprocessingMethod corr = Types::kNone,
- Types::SBType type = Types::kSignal );
+ Types::EPreprocessingMethod corr = Types::kNone,
+ Types::ESBType type = Types::kSignal );
// Create the search tree from the event collection
// using ONLY the variables specified in "theVars"
diff --git a/tmva/inc/BinarySearchTreeNode.h b/tmva/inc/BinarySearchTreeNode.h
index 8a2ba84e5ec..e41d366b4ad 100644
--- a/tmva/inc/BinarySearchTreeNode.h
+++ b/tmva/inc/BinarySearchTreeNode.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: BinarySearchTreeNode.h,v 1.4 2006/11/13 15:49:49 helgevoss Exp $
+// @(#)root/tmva $Id: BinarySearchTreeNode.h,v 1.5 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/BinaryTree.h b/tmva/inc/BinaryTree.h
index fd71c7077d9..0795c54862e 100644
--- a/tmva/inc/BinaryTree.h
+++ b/tmva/inc/BinaryTree.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: BinaryTree.h,v 1.24 2006/11/14 14:19:17 andreas.hoecker Exp $
+// @(#)root/tmva $Id: BinaryTree.h,v 1.25 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/CrossEntropy.h b/tmva/inc/CrossEntropy.h
index 7b78f04171a..36280439304 100644
--- a/tmva/inc/CrossEntropy.h
+++ b/tmva/inc/CrossEntropy.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: CrossEntropy.h,v 1.9 2006/11/06 00:10:16 helgevoss Exp $
+// @(#)root/tmva $Id: CrossEntropy.h,v 1.10 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -12,7 +12,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/inc/DataSet.h b/tmva/inc/DataSet.h
index beace90ab4e..ea0393fdc35 100644
--- a/tmva/inc/DataSet.h
+++ b/tmva/inc/DataSet.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: DataSet.h,v 1.39 2006/11/14 00:47:42 stelzer Exp $
+// @(#)root/tmva $Id: DataSet.h,v 1.41 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -82,8 +82,6 @@ namespace TMVA {
DataSet();
virtual ~DataSet();
- enum TreeType { kTraining=0, kTesting, kMaxTreeType };
-
const char* GetName() const { return "DataSet"; }
// the tree data
@@ -138,7 +136,7 @@ namespace TMVA {
// plot variables
// possible values for tree are 'training', 'multi'
- void PlotVariables( TString tree, TString folderName, Types::PreprocessingMethod corr = Types::kNone );
+ void PlotVariables( TString tree, TString folderName, Types::EPreprocessingMethod corr = Types::kNone );
// auxiliary functions to compute decorrelation
void GetCorrelationMatrix( Bool_t isSignal, TMatrixDBase* mat );
@@ -151,68 +149,68 @@ namespace TMVA {
// properties of the dataset
// normalisation init
- void CalcNorm(Types::PreprocessingMethod corr = Types::kNone);
+ void CalcNorm(Types::EPreprocessingMethod corr = Types::kNone);
// normalisation accessors
- Double_t GetRMS( Int_t ivar, Types::PreprocessingMethod corr = Types::kNone) const {
+ Double_t GetRMS( Int_t ivar, Types::EPreprocessingMethod corr = Types::kNone) const {
return fVariables[ivar].GetRMS(corr);
}
- Double_t GetRMS( const TString& var, Types::PreprocessingMethod corr = Types::kNone) const {
+ Double_t GetRMS( const TString& var, Types::EPreprocessingMethod corr = Types::kNone) const {
return GetRMS(FindVar(var), corr);
}
- Double_t GetMean( Int_t ivar, Types::PreprocessingMethod corr = Types::kNone) const {
+ Double_t GetMean( Int_t ivar, Types::EPreprocessingMethod corr = Types::kNone) const {
return fVariables[ivar].GetMean(corr);
}
- Double_t GetMean( const TString& var, Types::PreprocessingMethod corr = Types::kNone) const {
+ Double_t GetMean( const TString& var, Types::EPreprocessingMethod corr = Types::kNone) const {
return GetMean(FindVar(var), corr);
}
- Double_t GetXmin( Int_t ivar, Types::PreprocessingMethod corr = Types::kNone) const {
+ Double_t GetXmin( Int_t ivar, Types::EPreprocessingMethod corr = Types::kNone) const {
return fVariables[ivar].GetMin(corr);
}
- Double_t GetXmax( Int_t ivar, Types::PreprocessingMethod corr = Types::kNone) const {
+ Double_t GetXmax( Int_t ivar, Types::EPreprocessingMethod corr = Types::kNone) const {
return fVariables[ivar].GetMax(corr);
}
- Double_t GetXmin( const TString& var, Types::PreprocessingMethod corr = Types::kNone) const {
+ Double_t GetXmin( const TString& var, Types::EPreprocessingMethod corr = Types::kNone) const {
return GetXmin(FindVar(var), corr);
}
- Double_t GetXmax( const TString& var, Types::PreprocessingMethod corr = Types::kNone) const {
+ Double_t GetXmax( const TString& var, Types::EPreprocessingMethod corr = Types::kNone) const {
return GetXmax(FindVar(var), corr);
}
- void SetRMS ( const TString& var, Double_t x, Types::PreprocessingMethod corr = Types::kNone) {
+ void SetRMS ( const TString& var, Double_t x, Types::EPreprocessingMethod corr = Types::kNone) {
SetRMS(FindVar(var), x, corr);
}
- void SetRMS( Int_t ivar, Double_t x, Types::PreprocessingMethod corr = Types::kNone) {
+ void SetRMS( Int_t ivar, Double_t x, Types::EPreprocessingMethod corr = Types::kNone) {
fVariables[ivar].SetRMS(x, corr);
}
- void SetMean ( const TString& var, Double_t x, Types::PreprocessingMethod corr = Types::kNone) {
+ void SetMean ( const TString& var, Double_t x, Types::EPreprocessingMethod corr = Types::kNone) {
SetMean(FindVar(var), x, corr);
}
- void SetMean( Int_t ivar, Double_t x, Types::PreprocessingMethod corr = Types::kNone) {
+ void SetMean( Int_t ivar, Double_t x, Types::EPreprocessingMethod corr = Types::kNone) {
fVariables[ivar].SetMean(x, corr);
}
- void SetXmin( Int_t ivar, Double_t x, Types::PreprocessingMethod corr = Types::kNone) {
+ void SetXmin( Int_t ivar, Double_t x, Types::EPreprocessingMethod corr = Types::kNone) {
fVariables[ivar].SetMin(x, corr);
}
- void SetXmax( Int_t ivar, Double_t x, Types::PreprocessingMethod corr = Types::kNone) {
+ void SetXmax( Int_t ivar, Double_t x, Types::EPreprocessingMethod corr = Types::kNone) {
fVariables[ivar].SetMax(x, corr);
}
- void SetXmin( const TString& var, Double_t x, Types::PreprocessingMethod corr = Types::kNone) {
+ void SetXmin( const TString& var, Double_t x, Types::EPreprocessingMethod corr = Types::kNone) {
SetXmin(FindVar(var), x, corr);
}
- void SetXmax( const TString& var, Double_t x, Types::PreprocessingMethod corr = Types::kNone) {
+ void SetXmax( const TString& var, Double_t x, Types::EPreprocessingMethod corr = Types::kNone) {
SetXmax(FindVar(var), x, corr);
}
- void UpdateNorm ( Int_t ivar, Double_t x, Types::PreprocessingMethod corr = Types::kNone);
+ void UpdateNorm ( Int_t ivar, Double_t x, Types::EPreprocessingMethod corr = Types::kNone);
// event reading
- Bool_t ReadEvent(TTree* tr, UInt_t evidx, Types::PreprocessingMethod corr = Types::kNone,
- Types::SBType type = Types::kSignal) const;
- Bool_t ReadTrainingEvent( UInt_t evidx, Types::PreprocessingMethod corr = Types::kNone,
- Types::SBType type = Types::kSignal) const {
+ Bool_t ReadEvent(TTree* tr, UInt_t evidx, Types::EPreprocessingMethod corr = Types::kNone,
+ Types::ESBType type = Types::kSignal) const;
+ Bool_t ReadTrainingEvent( UInt_t evidx, Types::EPreprocessingMethod corr = Types::kNone,
+ Types::ESBType type = Types::kSignal) const {
return ReadEvent(GetTrainingTree(),evidx,corr,type);
}
- Bool_t ReadTestEvent( UInt_t evidx, Types::PreprocessingMethod corr = Types::kNone,
- Types::SBType type = Types::kSignal) const {
+ Bool_t ReadTestEvent( UInt_t evidx, Types::EPreprocessingMethod corr = Types::kNone,
+ Types::ESBType type = Types::kSignal) const {
return ReadEvent(GetTestTree(),evidx, corr,type);
}
@@ -229,23 +227,23 @@ namespace TMVA {
const TMVA::Event& Event() const { return *fEvent; } // Warning, this requires an existing event object
// decorrelation Matrix accessors
- const TMatrixD* CorrelationMatrix (Types::SBType sigbgd) const { return fDecorrMatrix[sigbgd]; }
- TPrincipal* PrincipalComponents(Types::SBType sigbgd) const { return fPrincipal[sigbgd]; }
+ const TMatrixD* CorrelationMatrix (Types::ESBType sigbgd) const { return fDecorrMatrix[sigbgd]; }
+ TPrincipal* PrincipalComponents(Types::ESBType sigbgd) const { return fPrincipal[sigbgd]; }
// the weight
void SetWeightExpression(const TString& expr) { fWeightExp = expr; }
// some dataset stats
- Int_t GetNEvtTrain() const { return fDataStats[kTraining][Types::kSBBoth]; }
- Int_t GetNEvtSigTrain() const { return fDataStats[kTraining][Types::kSignal]; }
- Int_t GetNEvtBkgdTrain() const { return fDataStats[kTraining][Types::kBackground]; }
- Int_t GetNEvtTest() const { return fDataStats[kTesting][Types::kSBBoth]; }
- Int_t GetNEvtSigTest() const { return fDataStats[kTesting][Types::kSignal]; }
- Int_t GetNEvtBkgdTest() const { return fDataStats[kTesting][Types::kBackground]; }
+ Int_t GetNEvtTrain() const { return fDataStats[Types::kTraining][Types::kSBBoth]; }
+ Int_t GetNEvtSigTrain() const { return fDataStats[Types::kTraining][Types::kSignal]; }
+ Int_t GetNEvtBkgdTrain() const { return fDataStats[Types::kTraining][Types::kBackground]; }
+ Int_t GetNEvtTest() const { return fDataStats[Types::kTesting][Types::kSBBoth]; }
+ Int_t GetNEvtSigTest() const { return fDataStats[Types::kTesting][Types::kSignal]; }
+ Int_t GetNEvtBkgdTest() const { return fDataStats[Types::kTesting][Types::kBackground]; }
// write and read functions
- void WriteVarsToStream ( std::ostream& o, Types::PreprocessingMethod corr ) const;
- void ReadVarsFromStream ( std::istream& istr, Types::PreprocessingMethod corr );
+ void WriteVarsToStream ( std::ostream& o, Types::EPreprocessingMethod corr ) const;
+ void ReadVarsFromStream ( std::istream& istr, Types::EPreprocessingMethod corr );
void WriteCorrMatToStream ( std::ostream& o ) const;
void ReadCorrMatFromStream( std::istream& istr );
@@ -254,29 +252,29 @@ namespace TMVA {
void ResetCurrentTree() { fCurrentTree = 0; }
// transformation for preprocessing
- Bool_t ApplyTransformation(Types::PreprocessingMethod corr = Types::kNone, Bool_t useSignal = kTRUE) const;
+ Bool_t ApplyTransformation(Types::EPreprocessingMethod corr = Types::kNone, Bool_t useSignal = kTRUE) const;
// preprocessing flag
Bool_t DoPreprocessing() const { return fDoPreprocessing; }
- Bool_t PreprocessingEnabled(Types::PreprocessingMethod corr) { return fPreprocessingEnabled[corr]; }
+ Bool_t PreprocessingEnabled(Types::EPreprocessingMethod corr) { return fPreprocessingEnabled[corr]; }
void SetPreprocessing( Bool_t doit ) {
fDoPreprocessing = doit;
fPreprocessingEnabled[Types::kDecorrelated] = doit;
}
- void EnablePreprocess( Types::PreprocessingMethod corr ) {
+ void EnablePreprocess( Types::EPreprocessingMethod corr ) {
fPreprocessingEnabled[corr] = kTRUE;
}
- Bool_t Preprocess(Types::PreprocessingMethod);
+ Bool_t Preprocess(Types::EPreprocessingMethod);
private:
// data manipulation helper functions
// helper functions for writing decorrelated data
- Bool_t PreparePreprocessing( Types::PreprocessingMethod corr, TTree* originalTree );
- Bool_t HasBeenPreprocessed( Types::PreprocessingMethod corr ) const {
+ Bool_t PreparePreprocessing( Types::EPreprocessingMethod corr, TTree* originalTree );
+ Bool_t HasBeenPreprocessed( Types::EPreprocessingMethod corr ) const {
return fFlagPreprocessed[corr];
}
- void FlagAsPreprocessed( Types::PreprocessingMethod corr ) {
+ void FlagAsPreprocessed( Types::EPreprocessingMethod corr ) {
fFlagPreprocessed[corr] = kTRUE;
}
@@ -295,7 +293,7 @@ namespace TMVA {
// plot variables
void PlotVariables( TTree* theTree, TString folderName = "input_variables",
- Types::PreprocessingMethod corr = Types::kDecorrelated );
+ Types::EPreprocessingMethod corr = Types::kDecorrelated );
// data members
@@ -320,7 +318,7 @@ namespace TMVA {
TTree* fMultiCutTestTree; //! tree used for testing of multicut method [correlated/decorrelated]
// data stats
- UInt_t fDataStats[kMaxTreeType][Types::kMaxSBType]; //! statistics of the dataset for training/test tree
+ UInt_t fDataStats[Types::kMaxTreeType][Types::kMaxSBType]; //! statistics of the dataset for training/test tree
//
TMatrixD* fCovarianceMatrix[2]; //! Covariance matrix [signal/background]
diff --git a/tmva/inc/DecisionTree.h b/tmva/inc/DecisionTree.h
index 23da170ba14..7e81c701589 100644
--- a/tmva/inc/DecisionTree.h
+++ b/tmva/inc/DecisionTree.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: DecisionTree.h,v 1.33 2006/11/14 14:19:17 andreas.hoecker Exp $
+// @(#)root/tmva $Id: DecisionTree.h,v 1.35 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -107,8 +107,8 @@ namespace TMVA {
void ClearTree();
// set pruning method
- enum PruneMethod { kExpectedErrorPruning=0, kCostComplexityPruning, kMCC };
- void SetPruneMethod( PruneMethod m = kExpectedErrorPruning ) { fPruneMethod = m; }
+ enum EPruneMethod { kExpectedErrorPruning=0, kCostComplexityPruning, kMCC };
+ void SetPruneMethod( EPruneMethod m = kExpectedErrorPruning ) { fPruneMethod = m; }
// recursive pruning of the tree
void PruneTree();
@@ -196,7 +196,7 @@ namespace TMVA {
Bool_t fUseSearchTree; //cut scan done with binary trees or simple event loop.
Double_t fPruneStrength; //a parameter to set the "amount" of pruning..needs to be adjusted
- PruneMethod fPruneMethod; // method used for prunig
+ EPruneMethod fPruneMethod; // method used for prunig
vector< Double_t > fVariableImportance; // the relative importance of the different variables
diff --git a/tmva/inc/DecisionTreeNode.h b/tmva/inc/DecisionTreeNode.h
index da731baf9e4..6a003fc5f38 100644
--- a/tmva/inc/DecisionTreeNode.h
+++ b/tmva/inc/DecisionTreeNode.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: DecisionTreeNode.h,v 1.21 2006/11/13 15:49:49 helgevoss Exp $
+// @(#)root/tmva $Id: DecisionTreeNode.h,v 1.22 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/Event.h b/tmva/inc/Event.h
index 6982eb15550..c77a6fcdb48 100644
--- a/tmva/inc/Event.h
+++ b/tmva/inc/Event.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Event.h,v 1.21 2006/11/14 14:19:17 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Event.h,v 1.22 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/Factory.h b/tmva/inc/Factory.h
index cd12cbe7e36..f9cc6cfd611 100644
--- a/tmva/inc/Factory.h
+++ b/tmva/inc/Factory.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Factory.h,v 1.25 2006/10/14 23:19:19 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Factory.h,v 1.27 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -16,13 +16,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -136,9 +136,9 @@ namespace TMVA {
void ProcessMultipleMVA();
Bool_t BookMethod( TString theMethodName, TString methodTitle, TString theOption = "" );
- Bool_t BookMethod( Types::MVA theMethod, TString methodTitle, TString theOption = "" );
- Bool_t BookMethod( TMVA::Types::MVA theMethod, TString methodTitle, TString methodOption,
- TMVA::Types::MVA theCommittee, TString committeeOption = "" );
+ Bool_t BookMethod( Types::EMVA theMethod, TString methodTitle, TString theOption = "" );
+ Bool_t BookMethod( TMVA::Types::EMVA theMethod, TString methodTitle, TString methodOption,
+ TMVA::Types::EMVA theCommittee, TString committeeOption = "" );
// booking the method with a given weight file --> testing or application only
// Bool_t BookMethod( IMethod *theMethod );
diff --git a/tmva/inc/GeneticANN.h b/tmva/inc/GeneticANN.h
index 87462786303..83fa2b7003e 100644
--- a/tmva/inc/GeneticANN.h
+++ b/tmva/inc/GeneticANN.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticANN.h,v 1.3 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticANN.h,v 1.4 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Matt Jachowski, Peter Speckmayer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -14,13 +14,13 @@
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Matt Jachowski <jachowski@stanford.edu> - Stanford University, USA *
* Peter Speckmayer <speckmay@mail.cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/GeneticBase.h b/tmva/inc/GeneticBase.h
index 2d933adbe74..b9379477d38 100644
--- a/tmva/inc/GeneticBase.h
+++ b/tmva/inc/GeneticBase.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticBase.h,v 1.14 2006/10/15 23:32:38 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticBase.h,v 1.15 2006/11/16 22:51:58 helgevoss Exp $
// Author: Peter Speckmayer
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/GeneticCuts.h b/tmva/inc/GeneticCuts.h
index c62c248d73b..3ca8b30e329 100644
--- a/tmva/inc/GeneticCuts.h
+++ b/tmva/inc/GeneticCuts.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticCuts.h,v 1.13 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticCuts.h,v 1.14 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Matt Jachowski, Peter Speckmayer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -14,13 +14,13 @@
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Matt Jachowski <jachowski@stanford.edu> - Stanford University, USA *
* Peter Speckmayer <speckmay@mail.cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/GeneticGenes.h b/tmva/inc/GeneticGenes.h
index 51ba164ce51..b9ab0449f07 100644
--- a/tmva/inc/GeneticGenes.h
+++ b/tmva/inc/GeneticGenes.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticGenes.h,v 1.8 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticGenes.h,v 1.9 2006/11/16 22:51:58 helgevoss Exp $
// Author: Peter Speckmayer
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/GeneticPopulation.h b/tmva/inc/GeneticPopulation.h
index fa6f48af0a6..cbb3c749f9d 100644
--- a/tmva/inc/GeneticPopulation.h
+++ b/tmva/inc/GeneticPopulation.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticPopulation.h,v 1.11 2006/10/15 23:32:38 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticPopulation.h,v 1.12 2006/11/16 22:51:58 helgevoss Exp $
// Author: Peter Speckmayer
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/GeneticRange.h b/tmva/inc/GeneticRange.h
index a5b11acb361..99a09043d91 100644
--- a/tmva/inc/GeneticRange.h
+++ b/tmva/inc/GeneticRange.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticRange.h,v 1.7 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticRange.h,v 1.8 2006/11/16 22:51:58 helgevoss Exp $
// Author: Peter Speckmayer
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/GiniIndex.h b/tmva/inc/GiniIndex.h
index 5a7ad2ed0bb..255d0c19cab 100644
--- a/tmva/inc/GiniIndex.h
+++ b/tmva/inc/GiniIndex.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GiniIndex.h,v 1.8 2006/11/06 00:10:17 helgevoss Exp $
+// @(#)root/tmva $Id: GiniIndex.h,v 1.9 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -23,7 +23,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/inc/IMethod.h b/tmva/inc/IMethod.h
index 0244880081e..08eefaf2740 100644
--- a/tmva/inc/IMethod.h
+++ b/tmva/inc/IMethod.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: IMethod.h,v 1.18 2006/11/02 15:44:49 andreas.hoecker Exp $
+// @(#)root/tmva $Id: IMethod.h,v 1.20 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -133,9 +133,9 @@ namespace TMVA {
virtual const TString& GetJobName ( void ) const = 0;
virtual const TString& GetMethodName ( void ) const = 0;
virtual const TString& GetMethodTitle( void ) const = 0;
- virtual const Types::MVA GetMethodType ( void ) const = 0;
+ virtual const Types::EMVA GetMethodType ( void ) const = 0;
- virtual Types::PreprocessingMethod GetPreprocessingMethod() const = 0;
+ virtual Types::EPreprocessingMethod GetPreprocessingMethod() const = 0;
virtual void SetJobName( TString jobName ) = 0;
diff --git a/tmva/inc/MethodBDT.h b/tmva/inc/MethodBDT.h
index 3dde936c4e3..ea580346290 100644
--- a/tmva/inc/MethodBDT.h
+++ b/tmva/inc/MethodBDT.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodBDT.h,v 1.33 2006/11/14 18:08:25 helgevoss Exp $
+// @(#)root/tmva $Id: MethodBDT.h,v 1.36 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -188,17 +188,19 @@ namespace TMVA {
void InitBDT( void );
//some histograms for monitoring
- TH1F* fBoostWeightHist; // weights applied in boosting
- TH2F* fErrFractHist; // error fraction vs tree number
- TTree* fMonitorNtuple; // monitoring ntuple
- Int_t fITree; // ntuple var: ith tree
- Double_t fBoostWeight; // ntuple var: boost weight
- Double_t fErrorFraction; // ntuple var: misclassification error fraction
- Int_t fNnodes; // ntuple var: nNodes
- Double_t fPruneStrength; // a parameter to set the "amount" of pruning..needs to be adjusted
- TMVA::DecisionTree::PruneMethod fPruneMethod; // method used for prunig
- TString fPruneMethodS; // prune method option String
- Bool_t fAutomatic; // use user given prune strength or automatically determined one using a validation sample
+ TH1F* fBoostWeightHist; // weights applied in boosting
+ TH1F* fBoostWeightVsTree;// weights applied in boosting vs tree number
+ TH1F* fErrFractHist; // error fraction vs tree number
+ TH1I* fNodesBeforePruningVsTree; // nNodesBeforePruning vs tree number
+ TH1I* fNodesAfterPruningVsTree; // nNodesAfterPruning vs tree number
+ TTree* fMonitorNtuple; // monitoring ntuple
+ Int_t fITree; // ntuple var: ith tree
+ Double_t fBoostWeight; // ntuple var: boost weight
+ Double_t fErrorFraction; // ntuple var: misclassification error fraction
+ Double_t fPruneStrength; // a parameter to set the "amount" of pruning..needs to be adjusted
+ TMVA::DecisionTree::EPruneMethod fPruneMethod; // method used for prunig
+ TString fPruneMethodS; // prune method option String
+ Bool_t fAutomatic; // use user given prune strength or automatically determined one using a validation sample
std::vector<Double_t> fVariableImportance; // the relative importance of the different variables
diff --git a/tmva/inc/MethodBase.h b/tmva/inc/MethodBase.h
index 58c6a5131e9..dc73e2044d2 100644
--- a/tmva/inc/MethodBase.h
+++ b/tmva/inc/MethodBase.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodBase.h,v 1.55 2006/11/14 23:02:57 stelzer Exp $
+// @(#)root/tmva $Id: MethodBase.h,v 1.57 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -76,7 +76,7 @@ namespace TMVA {
public:
- enum WeightFileType { kROOT=0, kTEXT };
+ enum EWeightFileType { kROOT=0, kTEXT };
// default constructur
MethodBase( TString jobName,
@@ -126,19 +126,19 @@ namespace TMVA {
const TString& GetMethodName ( void ) const { return fMethodName; }
const char* GetName ( void ) const { return GetMethodName().Data(); }
const TString& GetMethodTitle( void ) const { return fMethodTitle; }
- const Types::MVA GetMethodType ( void ) const { return fMethodType; }
+ const Types::EMVA GetMethodType ( void ) const { return fMethodType; }
void SetJobName ( TString jobName ) { fJobName = jobName; }
void SetMethodName ( TString methodName ) { fMethodName = methodName; }
void SetMethodTitle( TString methodTitle ) { fMethodTitle = methodTitle; }
- void SetMethodType ( Types::MVA methodType ) { fMethodType = methodType; }
+ void SetMethodType ( Types::EMVA methodType ) { fMethodType = methodType; }
TString GetOptions ( void ) const { return fOptions; }
TString GetWeightFileExtension( void ) const { return fFileExtension; }
void SetWeightFileExtension( TString fileExtension ) { fFileExtension = fileExtension; }
- void SetWeightFileType( WeightFileType w ) { fWeightFileType = w; }
- WeightFileType GetWeightFileType() const { return fWeightFileType; }
+ void SetWeightFileType( EWeightFileType w ) { fWeightFileType = w; }
+ EWeightFileType GetWeightFileType() const { return fWeightFileType; }
TString GetWeightFileDir( void ) const { return fFileDir; }
@@ -182,14 +182,14 @@ namespace TMVA {
virtual Double_t GetmuTransform ( TTree* );
// normalisation accessors
- Double_t GetXmin( Int_t ivar, Types::PreprocessingMethod corr = Types::kNone ) const { return fXminNorm[(Int_t) corr][ivar]; }
- Double_t GetXmax( Int_t ivar, Types::PreprocessingMethod corr = Types::kNone ) const { return fXmaxNorm[(Int_t) corr][ivar]; }
- Double_t GetXmin( const TString& var, Types::PreprocessingMethod corr = Types::kNone ) const { return GetXmin(Data().FindVar(var), corr); }
- Double_t GetXmax( const TString& var, Types::PreprocessingMethod corr = Types::kNone ) const { return GetXmax(Data().FindVar(var), corr); }
- void SetXmin( Int_t ivar, Double_t x, Types::PreprocessingMethod corr = Types::kNone ) { fXminNorm[(Int_t) corr][ivar] = x; }
- void SetXmax( Int_t ivar, Double_t x, Types::PreprocessingMethod corr = Types::kNone ) { fXmaxNorm[(Int_t) corr][ivar] = x; }
- void SetXmin( const TString& var, Double_t x, Types::PreprocessingMethod corr = Types::kNone ) { SetXmin(Data().FindVar(var), x, corr); }
- void SetXmax( const TString& var, Double_t x, Types::PreprocessingMethod corr = Types::kNone ) { SetXmax(Data().FindVar(var), x, corr); }
+ Double_t GetXmin( Int_t ivar, Types::EPreprocessingMethod corr = Types::kNone ) const { return fXminNorm[(Int_t) corr][ivar]; }
+ Double_t GetXmax( Int_t ivar, Types::EPreprocessingMethod corr = Types::kNone ) const { return fXmaxNorm[(Int_t) corr][ivar]; }
+ Double_t GetXmin( const TString& var, Types::EPreprocessingMethod corr = Types::kNone ) const { return GetXmin(Data().FindVar(var), corr); }
+ Double_t GetXmax( const TString& var, Types::EPreprocessingMethod corr = Types::kNone ) const { return GetXmax(Data().FindVar(var), corr); }
+ void SetXmin( Int_t ivar, Double_t x, Types::EPreprocessingMethod corr = Types::kNone ) { fXminNorm[(Int_t) corr][ivar] = x; }
+ void SetXmax( Int_t ivar, Double_t x, Types::EPreprocessingMethod corr = Types::kNone ) { fXmaxNorm[(Int_t) corr][ivar] = x; }
+ void SetXmin( const TString& var, Double_t x, Types::EPreprocessingMethod corr = Types::kNone ) { SetXmin(Data().FindVar(var), x, corr); }
+ void SetXmax( const TString& var, Double_t x, Types::EPreprocessingMethod corr = Types::kNone ) { SetXmax(Data().FindVar(var), x, corr); }
// main normalization method is in Tools
Double_t Norm ( Int_t ivar, Double_t x ) const;
@@ -202,18 +202,18 @@ namespace TMVA {
// write method-specific histograms to file
void WriteEvaluationHistosToFile( TDirectory* targetDir );
- Types::PreprocessingMethod GetPreprocessingMethod() const { return fPreprocessingMethod; }
- void SetPreprocessingMethod ( Types::PreprocessingMethod m ) { fPreprocessingMethod = m; }
+ Types::EPreprocessingMethod GetPreprocessingMethod() const { return fPreprocessingMethod; }
+ void SetPreprocessingMethod ( Types::EPreprocessingMethod m ) { fPreprocessingMethod = m; }
Bool_t Verbose( void ) const { return fVerbose; }
void SetVerbose( Bool_t v = kTRUE ) { fVerbose = v; }
DataSet& Data() const { return fData; }
- Bool_t ReadTrainingEvent( UInt_t ievt, Types::SBType type = Types::kMaxSBType ) {
+ Bool_t ReadTrainingEvent( UInt_t ievt, Types::ESBType type = Types::kMaxSBType ) {
return Data().ReadTrainingEvent( ievt, GetPreprocessingMethod(),
(type == Types::kMaxSBType) ? GetPreprocessingType() : type );
}
- virtual Bool_t ReadTestEvent( UInt_t ievt, Types::SBType type = Types::kMaxSBType ) {
+ virtual Bool_t ReadTestEvent( UInt_t ievt, Types::ESBType type = Types::kMaxSBType ) {
return Data().ReadTestEvent( ievt, GetPreprocessingMethod(),
(type == Types::kMaxSBType) ? GetPreprocessingType() : type );
}
@@ -233,8 +233,8 @@ namespace TMVA {
protected:
// used in efficiency computation
- enum CutOrientation { kNegative = -1, kPositive = +1 };
- CutOrientation GetCutOrientation() const { return fCutOrientation; }
+ enum ECutOrientation { kNegative = -1, kPositive = +1 };
+ ECutOrientation GetCutOrientation() const { return fCutOrientation; }
// reset required for RootFinder
void ResetThisBase( void ) { fgThisBase = this; }
@@ -245,17 +245,17 @@ namespace TMVA {
void SetSignalReferenceCut( Double_t cut ) { fSignalReferenceCut = cut; }
// some basic statistical analysis
- void Statistics( TMVA::Types::TreeType treeType, const TString& theVarName,
+ void Statistics( TMVA::Types::ETreeType treeType, const TString& theVarName,
Double_t&, Double_t&, Double_t&,
Double_t&, Double_t&, Double_t&, Bool_t norm = kFALSE );
- Types::SBType GetPreprocessingType() const { return fPreprocessingType; }
- void SetPreprocessingType( Types::SBType t ) { fPreprocessingType = t; }
+ Types::ESBType GetPreprocessingType() const { return fPreprocessingType; }
+ void SetPreprocessingType( Types::ESBType t ) { fPreprocessingType = t; }
private:
Double_t fSignalReferenceCut; // minimum requirement on the MVA output to declare an event signal-like
- Types::SBType fPreprocessingType; // this is the event type (sig or bgd) assumed for preprocessing
+ Types::ESBType fPreprocessingType; // this is the event type (sig or bgd) assumed for preprocessing
private:
@@ -291,7 +291,7 @@ namespace TMVA {
TString fJobName; // name of job -> user defined, appears in weight files
TString fMethodName; // name of the method (set in derived class)
- Types::MVA fMethodType; // type of method (set in derived class)
+ Types::EMVA fMethodType; // type of method (set in derived class)
TString fMethodTitle; // user-defined title for method (used for weight-file names)
TString fTestvar; // variable used in evaluation, etc (mostly the MVA)
TString fTestvarPrefix; // 'MVA_' prefix of MVA variable
@@ -309,7 +309,7 @@ namespace TMVA {
TString fFileDir; // unix sub-directory for weight files (default: "weights")
TString fWeightFile; // weight file name
- WeightFileType fWeightFileType; // The type of weight file {kROOT,kTEXT}
+ EWeightFileType fWeightFileType; // The type of weight file {kROOT,kTEXT}
protected:
@@ -362,7 +362,7 @@ namespace TMVA {
Double_t fXmax; // maximum (signal and background)
Bool_t fUseDecorr; // Use decorrelated Variables (kept for backward compatibility)
- Types::PreprocessingMethod fPreprocessingMethod; // Decorrelation, PCA, etc.
+ Types::EPreprocessingMethod fPreprocessingMethod; // Decorrelation, PCA, etc.
TString fPreprocessingString; // labels preprocessing method
TString fPreprocessingTypeString; // labels preprocessing type
@@ -378,7 +378,7 @@ namespace TMVA {
Int_t fNbinsH; // number of bins in evaluation histograms
// orientation of cut: depends on signal and background mean values
- CutOrientation fCutOrientation; // +1 if Sig>Bkg, -1 otherwise
+ ECutOrientation fCutOrientation; // +1 if Sig>Bkg, -1 otherwise
// for root finder
TSpline1* fSplRefS; // helper splines for RootFinder (signal)
diff --git a/tmva/inc/MethodBayesClassifier.h b/tmva/inc/MethodBayesClassifier.h
index 64456f6a9c9..0ee55d16f86 100644
--- a/tmva/inc/MethodBayesClassifier.h
+++ b/tmva/inc/MethodBayesClassifier.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodBayesClassifier.h,v 1.2 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodBayesClassifier.h,v 1.3 2006/11/16 22:51:58 helgevoss Exp $
// Author: Abhishek Narain
/**********************************************************************************
@@ -17,7 +17,7 @@
* University of Houston, *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/MethodCFMlpANN.h b/tmva/inc/MethodCFMlpANN.h
index 7ea8ea05fcb..a1e521a87a8 100644
--- a/tmva/inc/MethodCFMlpANN.h
+++ b/tmva/inc/MethodCFMlpANN.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCFMlpANN.h,v 1.20 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodCFMlpANN.h,v 1.21 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -57,13 +57,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -71,7 +71,7 @@
* (http://tmva.sourceforge.net/LICENSE) *
* *
* File and Version Information: *
- * $Id: MethodCFMlpANN.h,v 1.20 2006/11/02 15:44:50 andreas.hoecker Exp $
+ * $Id: MethodCFMlpANN.h,v 1.21 2006/11/16 22:51:58 helgevoss Exp $
**********************************************************************************/
#ifndef ROOT_TMVA_MethodCFMlpANN
diff --git a/tmva/inc/MethodCFMlpANN_Utils.h b/tmva/inc/MethodCFMlpANN_Utils.h
index 9c9d10ca224..2e02aa9971c 100644
--- a/tmva/inc/MethodCFMlpANN_Utils.h
+++ b/tmva/inc/MethodCFMlpANN_Utils.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCFMlpANN_Utils.h,v 1.13 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodCFMlpANN_Utils.h,v 1.14 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -18,13 +18,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/MethodCFMlpANN_def.h b/tmva/inc/MethodCFMlpANN_def.h
index f9db848bcff..880ae1e77a8 100644
--- a/tmva/inc/MethodCFMlpANN_def.h
+++ b/tmva/inc/MethodCFMlpANN_def.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCFMlpANN_def.h,v 1.5 2006/08/30 22:19:58 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodCFMlpANN_def.h,v 1.6 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/MethodCommittee.h b/tmva/inc/MethodCommittee.h
index 9c2cb54b848..216fec55bdf 100644
--- a/tmva/inc/MethodCommittee.h
+++ b/tmva/inc/MethodCommittee.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCommittee.h,v 1.5 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodCommittee.h,v 1.7 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -57,7 +57,7 @@ namespace TMVA {
TString committeeTitle,
DataSet& theData,
TString committeeOptions,
- Types::MVA method,
+ Types::EMVA method,
TString methodOptions,
TDirectory* theTargetDir = 0 );
@@ -122,7 +122,7 @@ namespace TMVA {
TString fBoostType; // string specifying the boost type
// options for the MVA method
- Types::MVA fMemberType; // the MVA method to be boosted
+ Types::EMVA fMemberType; // the MVA method to be boosted
TString fMemberOption; // the options for that method
Bool_t fUseMemberDecision; // use binary information from IsSignal
diff --git a/tmva/inc/MethodCuts.h b/tmva/inc/MethodCuts.h
index b35e6bb71d5..38960edf799 100644
--- a/tmva/inc/MethodCuts.h
+++ b/tmva/inc/MethodCuts.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCuts.h,v 1.34 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodCuts.h,v 1.36 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Matt Jachowski, Peter Speckmayer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -16,13 +16,13 @@
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Matt Jachowski <jachowski@stanford.edu> - Stanford University, USA *
* Peter Speckmayer <speckmay@mail.cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -123,34 +123,34 @@ namespace TMVA {
private:
// determines type of data to be optimised
- enum ConstrainType { kConstrainEffS = 0,
- kConstrainEffB } fConstrainType;
+ enum EConstrainType { kConstrainEffS = 0,
+ kConstrainEffB } fConstrainType;
// optimisation method
- enum FitMethodType { kUseMonteCarlo = 0,
- kUseGeneticAlgorithm,
- kUseSimulatedAnnealing };
+ enum EFitMethodType { kUseMonteCarlo = 0,
+ kUseGeneticAlgorithm,
+ kUseSimulatedAnnealing };
// efficiency calculation method
// - kUseEventSelection: computes efficiencies from given data sample
// - kUsePDFs : creates smoothed PDFs from data samples, and
// uses this to compute efficiencies
- enum EffMethod { kUseEventSelection = 0,
- kUsePDFs };
+ enum EEffMethod { kUseEventSelection = 0,
+ kUsePDFs };
// improve the Monte Carlo by providing some additional information
- enum FitParameters { kNotEnforced = 0,
- kForceMin,
- kForceMax,
- kForceSmart,
- kForceVerySmart };
+ enum EFitParameters { kNotEnforced = 0,
+ kForceMin,
+ kForceMax,
+ kForceSmart,
+ kForceVerySmart };
// general
TString fFitMethodS; // chosen fit method (string)
- FitMethodType fFitMethod; // chosen fit method
+ EFitMethodType fFitMethod; // chosen fit method
TString fEffMethodS; // chosen efficiency calculation method (string)
- EffMethod fEffMethod; // chosen efficiency calculation method
- vector<FitParameters>* fFitParams; // vector for series of fit methods
+ EEffMethod fEffMethod; // chosen efficiency calculation method
+ vector<EFitParameters>* fFitParams; // vector for series of fit methods
Double_t fTestSignalEff; // used to test optimized signal efficiency
Double_t fEffSMin; // used to test optimized signal efficiency
Double_t fEffSMax; // used to test optimized signal efficiency
diff --git a/tmva/inc/MethodFisher.h b/tmva/inc/MethodFisher.h
index f6ce6aca5f5..4772543b1d0 100644
--- a/tmva/inc/MethodFisher.h
+++ b/tmva/inc/MethodFisher.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodFisher.h,v 1.19 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodFisher.h,v 1.21 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Xavier Prudent, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -17,13 +17,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -31,7 +31,7 @@
* (http://tmva.sourceforge.net/LICENSE) *
* *
* File and Version Information: *
- * $Id: MethodFisher.h,v 1.19 2006/11/02 15:44:50 andreas.hoecker Exp $
+ * $Id: MethodFisher.h,v 1.21 2006/11/17 14:59:24 stelzer Exp $
**********************************************************************************/
#ifndef ROOT_TMVA_MethodFisher
@@ -90,8 +90,8 @@ namespace TMVA {
// calculate the MVA value
virtual Double_t GetMvaValue();
- enum FisherMethod { kFisher, kMahalanobis };
- virtual FisherMethod GetFisherMethod( void ) { return fFisherMethod; }
+ enum EFisherMethod { kFisher, kMahalanobis };
+ virtual EFisherMethod GetFisherMethod( void ) { return fFisherMethod; }
// ranking of input variables
const Ranking* CreateRanking();
@@ -146,7 +146,7 @@ namespace TMVA {
Double_t fF0;
// method to be used (Fisher or Mahalanobis)
- FisherMethod fFisherMethod;
+ EFisherMethod fFisherMethod;
// default initialisation called by all constructors
void InitFisher( void );
diff --git a/tmva/inc/MethodHMatrix.h b/tmva/inc/MethodHMatrix.h
index 5acb385b416..52f31b8f769 100644
--- a/tmva/inc/MethodHMatrix.h
+++ b/tmva/inc/MethodHMatrix.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodHMatrix.h,v 1.18 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodHMatrix.h,v 1.20 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Xavier Prudent, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -18,13 +18,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -95,8 +95,8 @@ namespace TMVA {
virtual void ProcessOptions();
// returns chi2 estimator for given type (signal or background)
- Double_t GetChi2( Event *e, Types::SBType ) const;
- Double_t GetChi2( Types::SBType ) const;
+ Double_t GetChi2( Event *e, Types::ESBType ) const;
+ Double_t GetChi2( Types::ESBType ) const;
// compute correlation matrices
void ComputeCovariance( Bool_t, TMatrixD* );
diff --git a/tmva/inc/MethodLikelihood.h b/tmva/inc/MethodLikelihood.h
index fdee69cfd71..26a2e55d70f 100644
--- a/tmva/inc/MethodLikelihood.h
+++ b/tmva/inc/MethodLikelihood.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodLikelihood.h,v 1.21 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodLikelihood.h,v 1.23 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -18,13 +18,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -96,7 +96,7 @@ namespace TMVA {
const Ranking* CreateRanking() { return 0; }
// overload test event reading
- virtual Bool_t ReadTestEvent(UInt_t ievt, Types::SBType type = Types::kSignal) {
+ virtual Bool_t ReadTestEvent(UInt_t ievt, Types::ESBType type = Types::kSignal) {
return Data().ReadTestEvent( ievt, Types::kNone, type );
}
@@ -113,7 +113,7 @@ namespace TMVA {
Int_t fAverageEvtPerBin; // average events per bin; used to calculate fNbins
// type of Splines used to smooth PDFs
- PDF::SmoothMethod fSmoothMethod;
+ PDF::ESmoothMethod fSmoothMethod;
// global weight file -- (needed !)
TFile* fFin;
diff --git a/tmva/inc/MethodMLP.h b/tmva/inc/MethodMLP.h
index 197092d5e2f..1299b703926 100644
--- a/tmva/inc/MethodMLP.h
+++ b/tmva/inc/MethodMLP.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodMLP.h,v 1.21 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodMLP.h,v 1.22 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Matt Jachowski
/**********************************************************************************
@@ -81,8 +81,8 @@ namespace TMVA {
// for GA
Double_t ComputeEstimator(const std::vector<Double_t>& parameters);
- enum TrainingMethod { kBP=0, kGA };
- enum BPTrainingMode { kSequential=0, kBatch };
+ enum ETrainingMethod { kBP=0, kGA };
+ enum EBPTrainingMode { kSequential=0, kBatch };
private:
@@ -96,7 +96,7 @@ namespace TMVA {
void InitializeLearningRates(); // although this is only needed by backprop
// used as a measure of success in all minimization techniques
- Double_t CalculateEstimator( Types::TreeType treeType = Types::kTrain );
+ Double_t CalculateEstimator( Types::ETreeType treeType = Types::kTraining );
// backpropagation functions
void BackPropagationMinimize( Int_t nEpochs );
@@ -125,23 +125,23 @@ namespace TMVA {
#endif
// general
- TrainingMethod fTrainingMethod; // method of training, BP or GA
- TString fTrainMethodS; // training method option param
+ ETrainingMethod fTrainingMethod; // method of training, BP or GA
+ TString fTrainMethodS; // training method option param
// backpropagation variables
- Double_t fLearnRate; // learning rate for synapse weight adjustments
- Double_t fDecayRate; // decay rate for above learning rate
- BPTrainingMode fBPMode; // backprop learning mode (sequential or batch)
- TString fBpModeS; // backprop learning mode option string (sequential or batch)
- Int_t fBatchSize; // batch size, only matters if in batch learning mode
- Int_t fTestRate; // test for overtraining performed at each #th epochs
+ Double_t fLearnRate; // learning rate for synapse weight adjustments
+ Double_t fDecayRate; // decay rate for above learning rate
+ EBPTrainingMode fBPMode; // backprop learning mode (sequential or batch)
+ TString fBpModeS; // backprop learning mode option string (sequential or batch)
+ Int_t fBatchSize; // batch size, only matters if in batch learning mode
+ Int_t fTestRate; // test for overtraining performed at each #th epochs
// genetic algorithm variables
- Int_t fGA_nsteps; // GA settings: number of steps
- Int_t fGA_preCalc; // GA settings: number of pre-calc steps
- Int_t fGA_SC_steps; // GA settings: SC_steps
- Int_t fGA_SC_offsteps; // GA settings: SC_offsteps
- Double_t fGA_SC_factor; // GA settings: SC_factor
+ Int_t fGA_nsteps; // GA settings: number of steps
+ Int_t fGA_preCalc; // GA settings: number of pre-calc steps
+ Int_t fGA_SC_steps; // GA settings: SC_steps
+ Int_t fGA_SC_offsteps; // GA settings: SC_offsteps
+ Double_t fGA_SC_factor; // GA settings: SC_factor
#ifdef MethodMLP_UseMinuit__
// minuit variables -- commented out because they rely on a static pointer
diff --git a/tmva/inc/MethodPDERS.h b/tmva/inc/MethodPDERS.h
index db7075d78c9..09a885ef849 100644
--- a/tmva/inc/MethodPDERS.h
+++ b/tmva/inc/MethodPDERS.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodPDERS.h,v 1.21 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodPDERS.h,v 1.23 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Yair Mahalalel, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -21,13 +21,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Yair Mahalalel <Yair.Mahalalel@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -132,7 +132,7 @@ namespace TMVA {
TString fVolumeRange; // option volume range
TString fKernelString; // option kernel estimator
- enum VolumeRangeMode {
+ enum EVolumeRangeMode {
kUnsupported = 0,
kMinMax,
kRMS,
@@ -140,7 +140,7 @@ namespace TMVA {
kUnscaled
} fVRangeMode;
- enum KernelEstimator {
+ enum EKernelEstimator {
kBox = 0,
kSphere,
kTeepee,
diff --git a/tmva/inc/MethodRuleFit.h b/tmva/inc/MethodRuleFit.h
index f879a51885f..c1be599c74a 100644
--- a/tmva/inc/MethodRuleFit.h
+++ b/tmva/inc/MethodRuleFit.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodRuleFit.h,v 1.24 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodRuleFit.h,v 1.25 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Fredrik Tegenfeldt, Helge Voss, Kai Voss
/**********************************************************************************
@@ -19,13 +19,13 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
* (http://tmva.sourceforge.net/LICENSE) *
- * $Id: MethodRuleFit.h,v 1.24 2006/11/02 15:44:50 andreas.hoecker Exp $
+ * $Id: MethodRuleFit.h,v 1.25 2006/11/16 22:51:58 helgevoss Exp $
**********************************************************************************/
#ifndef ROOT_TMVA_MethodRuleFit
diff --git a/tmva/inc/MethodSVM.h b/tmva/inc/MethodSVM.h
index de68da37dea..3f2b73499a0 100644
--- a/tmva/inc/MethodSVM.h
+++ b/tmva/inc/MethodSVM.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodSVM.h,v 1.11 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodSVM.h,v 1.12 2006/11/16 22:51:58 helgevoss Exp $
// Author: Marcin ....
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/MethodTMlpANN.h b/tmva/inc/MethodTMlpANN.h
index 9d11bd36861..ece6c887a5f 100644
--- a/tmva/inc/MethodTMlpANN.h
+++ b/tmva/inc/MethodTMlpANN.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodTMlpANN.h,v 1.19 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodTMlpANN.h,v 1.20 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -16,13 +16,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/MethodVariable.h b/tmva/inc/MethodVariable.h
index d57c73b415d..ef63a604e4e 100644
--- a/tmva/inc/MethodVariable.h
+++ b/tmva/inc/MethodVariable.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodVariable.h,v 1.15 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodVariable.h,v 1.16 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -14,13 +14,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/Methods.h b/tmva/inc/Methods.h
index 0f744f3c6e4..1234f692c13 100644
--- a/tmva/inc/Methods.h
+++ b/tmva/inc/Methods.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Methods.h,v 1.3 2006/10/26 19:55:40 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Methods.h,v 1.4 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/MisClassificationError.h b/tmva/inc/MisClassificationError.h
index 7f305190983..d68f8dd735a 100644
--- a/tmva/inc/MisClassificationError.h
+++ b/tmva/inc/MisClassificationError.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MisClassificationError.h,v 1.9 2006/11/06 00:10:17 helgevoss Exp $
+// @(#)root/tmva $Id: MisClassificationError.h,v 1.10 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -15,7 +15,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/inc/MsgLogger.h b/tmva/inc/MsgLogger.h
index c4477d94d81..63fc5a0a21b 100644
--- a/tmva/inc/MsgLogger.h
+++ b/tmva/inc/MsgLogger.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MsgLogger.h,v 1.6 2006/10/17 21:22:29 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MsgLogger.h,v 1.8 2006/11/16 22:51:59 helgevoss Exp $
// Author: Attila Krasznahorkay
/**********************************************************************************
@@ -15,7 +15,7 @@
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -48,7 +48,7 @@
namespace TMVA {
// define outside of class to facilite access
- enum MsgType {
+ enum EMsgType {
kVERBOSE = 1,
kDEBUG = 2,
kINFO = 3,
@@ -62,16 +62,16 @@ namespace TMVA {
public:
- MsgLogger( const TObject* source, MsgType minType = kINFO );
- MsgLogger( const std::string& source, MsgType minType = kINFO );
- MsgLogger( MsgType minType = kINFO );
+ MsgLogger( const TObject* source, EMsgType minType = kINFO );
+ MsgLogger( const std::string& source, EMsgType minType = kINFO );
+ MsgLogger( EMsgType minType = kINFO );
MsgLogger( const MsgLogger& parent );
~MsgLogger();
// Accessors
void SetSource ( const std::string& source ) { fStrSource = source; }
- MsgType GetMinType() const { return fMinType; }
- void SetMinType( MsgType minType ) { fMinType = minType; }
+ EMsgType GetMinType() const { return fMinType; }
+ void SetMinType( EMsgType minType ) { fMinType = minType; }
UInt_t GetMaxSourceSize() const { return (UInt_t)fMaxSourceSize; }
std::string GetPrintedSource() const;
std::string GetFormattedSource() const;
@@ -80,7 +80,7 @@ namespace TMVA {
MsgLogger& operator= ( const MsgLogger& parent );
// Stream modifier(s)
- static MsgLogger& endmsg( MsgLogger& logger );
+ static MsgLogger& Endmsg( MsgLogger& logger );
// Accept stream modifiers
MsgLogger& operator<< ( MsgLogger& ( *_f )( MsgLogger& ) );
@@ -88,7 +88,7 @@ namespace TMVA {
MsgLogger& operator<< ( std::ios& ( *_f )( std::ios& ) );
// Accept message type specification
- MsgLogger& operator<< ( MsgType type );
+ MsgLogger& operator<< ( EMsgType type );
// For all the "conventional" inputs
template <class T> MsgLogger& operator<< ( T arg ) {
@@ -101,18 +101,18 @@ namespace TMVA {
// private utility routines
void Send();
void InitMaps();
- void WriteMsg( MsgType type, const std::string& line ) const;
+ void WriteMsg( EMsgType type, const std::string& line ) const;
- const TObject* fObjSource; // the source TObject (used for name)
- std::string fStrSource; // alternative string source
- const std::string fPrefix; // the prefix of the source name
- const std::string fSuffix; // suffix following source name
- MsgType fActiveType; // active type
- const std::string::size_type fMaxSourceSize; // maximum length of source name
+ const TObject* fObjSource; // the source TObject (used for name)
+ std::string fStrSource; // alternative string source
+ const std::string fPrefix; // the prefix of the source name
+ const std::string fSuffix; // suffix following source name
+ EMsgType fActiveType; // active type
+ const std::string::size_type fMaxSourceSize; // maximum length of source name
- std::map<MsgType, std::string> fTypeMap; // matches output types with strings
- std::map<MsgType, std::string> fColorMap; // matches output types with terminal colors
- MsgType fMinType; // minimum type for output
+ std::map<EMsgType, std::string> fTypeMap; // matches output types with strings
+ std::map<EMsgType, std::string> fColorMap; // matches output types with terminal colors
+ EMsgType fMinType; // minimum type for output
ClassDef(MsgLogger,0) // ostringstream derivative to redirect and format logging output
;
@@ -135,7 +135,7 @@ namespace TMVA {
return *this;
}
- inline MsgLogger& MsgLogger::operator<< ( MsgType type )
+ inline MsgLogger& MsgLogger::operator<< ( EMsgType type )
{
fActiveType = type;
return *this;
@@ -144,8 +144,8 @@ namespace TMVA {
// Although the proper definition of "Endl" as a function pointer
// would be nicer C++-wise, it introduces some "unused variable"
// warnings so let's use the #define definition after all...
- // static MsgLogger& ( *Endl )( MsgLogger& ) = &MsgLogger::endmsg;
-#define Endl MsgLogger::endmsg
+ // static MsgLogger& ( *Endl )( MsgLogger& ) = &MsgLogger::Endmsg;
+#define Endl MsgLogger::Endmsg
}
diff --git a/tmva/inc/Node.h b/tmva/inc/Node.h
index 59abacb9cdc..e559c19c737 100644
--- a/tmva/inc/Node.h
+++ b/tmva/inc/Node.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Node.h,v 1.19 2006/11/13 15:49:49 helgevoss Exp $
+// @(#)root/tmva $Id: Node.h,v 1.20 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/Option.h b/tmva/inc/Option.h
index aa2ab79946e..c1d513e76be 100644
--- a/tmva/inc/Option.h
+++ b/tmva/inc/Option.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Option.h,v 1.17 2006/11/14 23:02:57 stelzer Exp $
+// @(#)root/tmva $Id: Option.h,v 1.18 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/PDF.h b/tmva/inc/PDF.h
index ffc2844a650..8962a8a8b64 100644
--- a/tmva/inc/PDF.h
+++ b/tmva/inc/PDF.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: PDF.h,v 1.18 2006/10/15 12:06:32 andreas.hoecker Exp $
+// @(#)root/tmva $Id: PDF.h,v 1.20 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -52,10 +52,10 @@ namespace TMVA {
public:
- enum SmoothMethod { kSpline0, kSpline1, kSpline2, kSpline3, kSpline5 };
+ enum ESmoothMethod { kSpline0, kSpline1, kSpline2, kSpline3, kSpline5 };
PDF( const TH1* theHist,
- PDF::SmoothMethod method = kSpline2,
+ ESmoothMethod method = kSpline2,
Int_t nsmooth = 0 );
virtual ~PDF( void );
diff --git a/tmva/inc/Ranking.h b/tmva/inc/Ranking.h
index ab920200e77..6df05e19f37 100644
--- a/tmva/inc/Ranking.h
+++ b/tmva/inc/Ranking.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Ranking.h,v 1.8 2006/10/15 23:32:38 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Ranking.h,v 1.9 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,18 +13,18 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
* (http://tmva.sourceforge.net/LICENSE) *
* *
* File and Version Information: *
- * $Id: Ranking.h,v 1.8 2006/10/15 23:32:38 andreas.hoecker Exp $
+ * $Id: Ranking.h,v 1.9 2006/11/16 22:51:59 helgevoss Exp $
**********************************************************************************/
#ifndef ROOT_TMVA_Ranking
diff --git a/tmva/inc/Reader.h b/tmva/inc/Reader.h
index 286875abcb4..85b71903334 100644
--- a/tmva/inc/Reader.h
+++ b/tmva/inc/Reader.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Reader.h,v 1.14 2006/11/13 23:43:34 stelzer Exp $
+// @(#)root/tmva $Id: Reader.h,v 1.16 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -14,13 +14,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -98,7 +98,7 @@ namespace TMVA {
private:
// this booking method is internal
- IMethod* BookMVA( Types::MVA method, TString weightfile );
+ IMethod* BookMVA( Types::EMVA method, TString weightfile );
DataSet * fDataSet; // the data set
diff --git a/tmva/inc/RootFinder.h b/tmva/inc/RootFinder.h
index 42bf50f6615..e7960a84f37 100644
--- a/tmva/inc/RootFinder.h
+++ b/tmva/inc/RootFinder.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: RootFinder.h,v 1.12 2006/10/15 22:34:22 andreas.hoecker Exp $
+// @(#)root/tmva $Id: RootFinder.h,v 1.13 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -14,13 +14,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/Rule.h b/tmva/inc/Rule.h
index f37070d6974..0bab346c7a1 100644
--- a/tmva/inc/Rule.h
+++ b/tmva/inc/Rule.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Rule.h,v 1.22 2006/10/23 01:51:39 stelzer Exp $
+// @(#)root/tmva $Id: Rule.h,v 1.23 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -21,7 +21,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
diff --git a/tmva/inc/RuleEnsemble.h b/tmva/inc/RuleEnsemble.h
index 418a40e0a6b..949dc29aa8e 100644
--- a/tmva/inc/RuleEnsemble.h
+++ b/tmva/inc/RuleEnsemble.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: RuleEnsemble.h,v 1.24 2006/10/23 01:51:39 stelzer Exp $
+// @(#)root/tmva $Id: RuleEnsemble.h,v 1.26 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -19,7 +19,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -60,7 +60,7 @@ namespace TMVA {
public:
- enum LearningModel { kFull, kRules, kLinear };
+ enum ELearningModel { kFull, kRules, kLinear };
// main constructor
RuleEnsemble( RuleFit* rf );
@@ -196,7 +196,7 @@ namespace TMVA {
inline const Bool_t DoLinear() const { return (fLearningModel==kFull) || (fLearningModel==kLinear); }
inline const Bool_t DoRules() const { return (fLearningModel==kFull) || (fLearningModel==kRules); }
const Bool_t DoFull() const { return (fLearningModel==kFull); }
- const LearningModel GetLearningModel() const { return fLearningModel; }
+ const ELearningModel GetLearningModel() const { return fLearningModel; }
const Double_t GetImportanceCut() const { return fImportanceCut; }
const Double_t GetOffset() const { return fOffset; }
const UInt_t GetNRules() const { return fRules.size(); }
@@ -262,7 +262,7 @@ namespace TMVA {
// evaluate linear terms used to fill fEventLinearVal
Double_t EvalLinEventRaw( UInt_t vind, const Event &e );
- LearningModel fLearningModel; // can be full (rules+linear), rules, linear
+ ELearningModel fLearningModel; // can be full (rules+linear), rules, linear
Double_t fImportanceCut; // minimum importance accepted
Double_t fOffset; // offset in discriminator function
std::vector< Rule* > fRules; // vector of rules
diff --git a/tmva/inc/RuleFit.h b/tmva/inc/RuleFit.h
index b0b126d75f6..930ae540554 100644
--- a/tmva/inc/RuleFit.h
+++ b/tmva/inc/RuleFit.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: RuleFit.h,v 1.24 2006/10/17 07:44:57 tegen Exp $
+// @(#)root/tmva $Id: RuleFit.h,v 1.25 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -17,7 +17,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
diff --git a/tmva/inc/RuleFitParams.h b/tmva/inc/RuleFitParams.h
index fbf4f2c8ed5..fd70b5c8e95 100644
--- a/tmva/inc/RuleFitParams.h
+++ b/tmva/inc/RuleFitParams.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: RuleFitParams.h,v 1.19 2006/11/14 14:23:32 andreas.hoecker Exp $
+// @(#)root/tmva $Id: RuleFitParams.h,v 1.20 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -23,7 +23,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
diff --git a/tmva/inc/SdivSqrtSplusB.h b/tmva/inc/SdivSqrtSplusB.h
index 70eae54e71b..92c2ceacf12 100644
--- a/tmva/inc/SdivSqrtSplusB.h
+++ b/tmva/inc/SdivSqrtSplusB.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: SdivSqrtSplusB.h,v 1.9 2006/11/06 00:10:17 helgevoss Exp $
+// @(#)root/tmva $Id: SdivSqrtSplusB.h,v 1.10 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,7 +13,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/inc/SeparationBase.h b/tmva/inc/SeparationBase.h
index 58b52b0a1d8..59650c4fc09 100644
--- a/tmva/inc/SeparationBase.h
+++ b/tmva/inc/SeparationBase.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: SeparationBase.h,v 1.11 2006/11/06 00:10:17 helgevoss Exp $
+// @(#)root/tmva $Id: SeparationBase.h,v 1.12 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -30,7 +30,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/inc/SimulatedAnnealingBase.h b/tmva/inc/SimulatedAnnealingBase.h
index 4ba0263be73..23381da02e2 100644
--- a/tmva/inc/SimulatedAnnealingBase.h
+++ b/tmva/inc/SimulatedAnnealingBase.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: SimulatedAnnealingBase.h,v 1.6 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: SimulatedAnnealingBase.h,v 1.7 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/SimulatedAnnealingCuts.h b/tmva/inc/SimulatedAnnealingCuts.h
index 7792ac8551e..d549e075817 100644
--- a/tmva/inc/SimulatedAnnealingCuts.h
+++ b/tmva/inc/SimulatedAnnealingCuts.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: SimulatedAnnealingCuts.h,v 1.5 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: SimulatedAnnealingCuts.h,v 1.6 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/TActivationChooser.h b/tmva/inc/TActivationChooser.h
index 5f4dd288a03..eedc2351fde 100644
--- a/tmva/inc/TActivationChooser.h
+++ b/tmva/inc/TActivationChooser.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: TActivationChooser.h,v 1.8 2006/10/17 21:22:30 andreas.hoecker Exp $
+// @(#)root/tmva $Id: TActivationChooser.h,v 1.10 2006/11/17 14:59:24 stelzer Exp $
// Author: Matt Jachowski
/**********************************************************************************
@@ -64,6 +64,8 @@ namespace TMVA {
TActivationChooser()
: fLogger( "TActivationChooser" )
{
+ // defaut constructor
+
fLINEAR = "linear";
fSIGMOID = "sigmoid";
fTANH = "tanh";
@@ -71,14 +73,17 @@ namespace TMVA {
}
virtual ~TActivationChooser() {}
- enum ActivationType { kLinear = 0,
- kSigmoid,
- kTanh,
- kRadial
+ enum EActivationType { kLinear = 0,
+ kSigmoid,
+ kTanh,
+ kRadial
};
- TActivation* CreateActivation(const ActivationType type) const
+ TActivation* CreateActivation(const EActivationType type) const
{
+ // instantiate the correct activation object according to the
+ // type choosen (given as the enumeration type)
+
switch (type) {
case kLinear: return new TActivationIdentity();
case kSigmoid: return new TActivationSigmoid();
@@ -93,6 +98,9 @@ namespace TMVA {
TActivation* CreateActivation(const TString type) const
{
+ // instantiate the correct activation object according to the
+ // type choosen (given by a TString)
+
if (type == fLINEAR) return CreateActivation(kLinear);
else if (type == fSIGMOID) return CreateActivation(kSigmoid);
else if (type == fTANH) return CreateActivation(kTanh);
@@ -105,6 +113,8 @@ namespace TMVA {
vector<TString>* GetAllActivationNames() const
{
+ // retuns the names of all know activation functions
+
vector<TString>* names = new vector<TString>();
names->push_back(fLINEAR);
names->push_back(fSIGMOID);
diff --git a/tmva/inc/TNeuron.h b/tmva/inc/TNeuron.h
index 06f1244058d..a05af06bfb9 100644
--- a/tmva/inc/TNeuron.h
+++ b/tmva/inc/TNeuron.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: TNeuron.h,v 1.16 2006/10/15 12:06:33 andreas.hoecker Exp $
+// @(#)root/tmva $Id: TNeuron.h,v 1.17 2006/11/16 19:42:44 stelzer Exp $
// Author: Matt Jachowski
/**********************************************************************************
@@ -126,7 +126,7 @@ namespace TMVA {
void InitNeuron();
void DeleteLinksArray( TObjArray*& links );
void PrintLinks ( TObjArray* links );
- void PrintMessage ( MsgType, TString message );
+ void PrintMessage ( EMsgType, TString message );
// inlined helper functions
Int_t NumLinks(TObjArray* links) {
diff --git a/tmva/inc/TNeuronInputChooser.h b/tmva/inc/TNeuronInputChooser.h
index 9483cf77d86..f6d505a3a28 100644
--- a/tmva/inc/TNeuronInputChooser.h
+++ b/tmva/inc/TNeuronInputChooser.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: TNeuronInputChooser.h,v 1.4 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: TNeuronInputChooser.h,v 1.5 2006/11/17 14:59:24 stelzer Exp $
// Author: Matt Jachowski
/**********************************************************************************
@@ -65,12 +65,12 @@ namespace TMVA {
}
virtual ~TNeuronInputChooser() {}
- enum NeuronInputType { kSum = 0,
- kSqSum,
- kAbsSum
+ enum ENeuronInputType { kSum = 0,
+ kSqSum,
+ kAbsSum
};
- TNeuronInput* CreateNeuronInput(const NeuronInputType type) const
+ TNeuronInput* CreateNeuronInput(const ENeuronInputType type) const
{
switch (type) {
case kSum: return new TNeuronInputSum();
diff --git a/tmva/inc/TSpline1.h b/tmva/inc/TSpline1.h
index 7e8dee4b73d..9797e7dc62e 100644
--- a/tmva/inc/TSpline1.h
+++ b/tmva/inc/TSpline1.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: TSpline1.h,v 1.9 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: TSpline1.h,v 1.10 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/TSpline2.h b/tmva/inc/TSpline2.h
index 3b0425d4f7b..0db07416c8b 100644
--- a/tmva/inc/TSpline2.h
+++ b/tmva/inc/TSpline2.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: TSpline2.h,v 1.10 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: TSpline2.h,v 1.11 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/Timer.h b/tmva/inc/Timer.h
index 554a4c14cf6..1e249966092 100644
--- a/tmva/inc/Timer.h
+++ b/tmva/inc/Timer.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Timer.h,v 1.9 2006/10/14 00:57:42 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Timer.h,v 1.10 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/Tools.h b/tmva/inc/Tools.h
index 77ba572aec7..33bbcaf0996 100644
--- a/tmva/inc/Tools.h
+++ b/tmva/inc/Tools.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Tools.h,v 1.26 2006/10/15 22:34:22 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Tools.h,v 1.27 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/inc/Types.h b/tmva/inc/Types.h
index a82a70a3a75..87f6250958e 100644
--- a/tmva/inc/Types.h
+++ b/tmva/inc/Types.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Types.h,v 1.25 2006/11/13 23:43:34 stelzer Exp $
+// @(#)root/tmva $Id: Types.h,v 1.28 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -54,32 +54,32 @@ namespace TMVA {
public:
// available MVA methods in TMVA
- enum MVA {
- Variable = 0,
- Cuts ,
- Likelihood ,
- PDERS ,
- HMatrix ,
- Fisher ,
- CFMlpANN ,
- TMlpANN ,
- BDT ,
- RuleFit ,
- SVM ,
- MLP ,
- BayesClassifier,
- Committee ,
+ enum EMVA {
+ kVariable = 0,
+ kCuts ,
+ kLikelihood ,
+ kPDERS ,
+ kHMatrix ,
+ kFisher ,
+ kCFMlpANN ,
+ kTMlpANN ,
+ kBDT ,
+ kRuleFit ,
+ kSVM ,
+ kMLP ,
+ kBayesClassifier,
+ kCommittee ,
kMaxMethod
};
- enum PreprocessingMethod {
+ enum EPreprocessingMethod {
kNone = 0,
kDecorrelated,
kPCA,
kMaxPreprocessingMethod
};
- enum SBType {
+ enum ESBType {
kSignal = 0,
kBackground,
kSBBoth,
@@ -87,18 +87,18 @@ namespace TMVA {
kTrueType
};
- enum TreeType { kTrain = 1, kTest };
+ enum ETreeType { kTraining = 0, kTesting, kMaxTreeType };
public:
static Types& Instance() { return fgTypesPtr ? *fgTypesPtr : *(fgTypesPtr = new Types()); }
~Types() {}
- const MVA GetMethodType( const TString& method ) const {
- std::map<TString, MVA>::const_iterator it = fStr2type.find( method );
+ const EMVA GetMethodType( const TString& method ) const {
+ std::map<TString, EMVA>::const_iterator it = fStr2type.find( method );
if (it == fStr2type.end()) {
fLogger << kFATAL << "unknown method in map: " << method << Endl;
- return Variable; // Inserted to get rid of GCC warning...
+ return kVariable; // Inserted to get rid of GCC warning...
}
else return it->second;
}
@@ -111,7 +111,7 @@ namespace TMVA {
private:
- std::map<TString, MVA> fStr2type; // types-to-text map
+ std::map<TString, EMVA> fStr2type; // types-to-text map
mutable MsgLogger fLogger; // message logger
};
diff --git a/tmva/inc/VariableInfo.h b/tmva/inc/VariableInfo.h
index b723104391a..5cc51ddc45d 100644
--- a/tmva/inc/VariableInfo.h
+++ b/tmva/inc/VariableInfo.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: VariableInfo.h,v 1.13 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: VariableInfo.h,v 1.15 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2006: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -60,22 +60,22 @@ namespace TMVA {
const TString& GetInternalVarName() const { return fInternalVarName; }
char VarType() const { return fVarType; }
char VarTypeOriginal() const { return fVarTypeOriginal; }
- Double_t GetMin (Types::PreprocessingMethod corr = Types::kNone) const { return fXminNorm[(Int_t) corr]; }
- Double_t GetMax (Types::PreprocessingMethod corr = Types::kNone) const { return fXmaxNorm[(Int_t) corr]; }
- Double_t GetMean(Types::PreprocessingMethod corr = Types::kNone) const { return fXmeanNorm[(Int_t) corr]; }
- Double_t GetRMS (Types::PreprocessingMethod corr = Types::kNone) const { return fXrmsNorm[(Int_t) corr]; }
+ Double_t GetMin (Types::EPreprocessingMethod corr = Types::kNone) const { return fXminNorm[(Int_t) corr]; }
+ Double_t GetMax (Types::EPreprocessingMethod corr = Types::kNone) const { return fXmaxNorm[(Int_t) corr]; }
+ Double_t GetMean(Types::EPreprocessingMethod corr = Types::kNone) const { return fXmeanNorm[(Int_t) corr]; }
+ Double_t GetRMS (Types::EPreprocessingMethod corr = Types::kNone) const { return fXrmsNorm[(Int_t) corr]; }
void SetExpression(const TString& s) { fExpression = s; }
void SetInternalVarName(const TString& s) { fInternalVarName = s; }
void SetVarType(char c) { fVarType = c; }
- void SetMin (Double_t v, Types::PreprocessingMethod corr = Types::kNone) { fXminNorm[(Int_t) corr] = v; }
- void SetMax (Double_t v, Types::PreprocessingMethod corr = Types::kNone) { fXmaxNorm[(Int_t) corr] = v; }
- void SetMean(Double_t v, Types::PreprocessingMethod corr = Types::kNone) { fXmeanNorm[(Int_t) corr] = v; }
- void SetRMS (Double_t v, Types::PreprocessingMethod corr = Types::kNone) { fXrmsNorm[(Int_t) corr] = v; }
+ void SetMin (Double_t v, Types::EPreprocessingMethod corr = Types::kNone) { fXminNorm[(Int_t) corr] = v; }
+ void SetMax (Double_t v, Types::EPreprocessingMethod corr = Types::kNone) { fXmaxNorm[(Int_t) corr] = v; }
+ void SetMean(Double_t v, Types::EPreprocessingMethod corr = Types::kNone) { fXmeanNorm[(Int_t) corr] = v; }
+ void SetRMS (Double_t v, Types::EPreprocessingMethod corr = Types::kNone) { fXrmsNorm[(Int_t) corr] = v; }
void SetExternalLink(void* p) { fExternalData = p; }
- void WriteToStream(std::ostream& o, Types::PreprocessingMethod corr) const;
- void ReadFromStream(std::istream& istr, Types::PreprocessingMethod corr);
+ void WriteToStream(std::ostream& o, Types::EPreprocessingMethod corr) const;
+ void ReadFromStream(std::istream& istr, Types::EPreprocessingMethod corr);
void* GetExternalLink() const { return fExternalData; }
// assignment operator (does not copy external link)
diff --git a/tmva/inc/Volume.h b/tmva/inc/Volume.h
index f9bdaf15295..b0130bcc95d 100644
--- a/tmva/inc/Volume.h
+++ b/tmva/inc/Volume.h
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Volume.h,v 1.10 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Volume.h,v 1.11 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/BinarySearchTree.cxx b/tmva/src/BinarySearchTree.cxx
index d88eef7c404..908fa5fac6d 100644
--- a/tmva/src/BinarySearchTree.cxx
+++ b/tmva/src/BinarySearchTree.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: BinarySearchTree.cxx,v 1.22 2006/11/06 00:10:12 helgevoss Exp $
+// @(#)root/tmva $Id: BinarySearchTree.cxx,v 1.24 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -27,7 +27,7 @@
* (http://tmva.sourceforge.net/LICENSE) *
* *
* File and Version Information: *
- * $Id: BinarySearchTree.cxx,v 1.22 2006/11/06 00:10:12 helgevoss Exp $
+ * $Id: BinarySearchTree.cxx,v 1.24 2006/11/17 14:59:23 stelzer Exp $
**********************************************************************************/
//_______________________________________________________________________
@@ -198,7 +198,7 @@ Double_t TMVA::BinarySearchTree::GetSumOfWeights( void ) const
//_______________________________________________________________________
Int_t TMVA::BinarySearchTree::Fill( const DataSet& ds, TTree* theTree, Int_t theType,
- Types::PreprocessingMethod corr, Types::SBType type )
+ Types::EPreprocessingMethod corr, Types::ESBType type )
{
// create the search tree from the events in the DataSet
Int_t nevents=0;
diff --git a/tmva/src/BinarySearchTreeNode.cxx b/tmva/src/BinarySearchTreeNode.cxx
index b2b03cee4ef..1216084ad7f 100644
--- a/tmva/src/BinarySearchTreeNode.cxx
+++ b/tmva/src/BinarySearchTreeNode.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: BinarySearchTreeNode.cxx,v 1.4 2006/11/13 15:49:49 helgevoss Exp $
+// @(#)root/tmva $Id: BinarySearchTreeNode.cxx,v 1.5 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* CopyRight (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/BinaryTree.cxx b/tmva/src/BinaryTree.cxx
index 82c3d4aff01..82715204586 100644
--- a/tmva/src/BinaryTree.cxx
+++ b/tmva/src/BinaryTree.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: BinaryTree.cxx,v 1.23 2006/11/13 15:49:49 helgevoss Exp $
+// @(#)root/tmva $Id: BinaryTree.cxx,v 1.24 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/CrossEntropy.cxx b/tmva/src/CrossEntropy.cxx
index 42d1325cab2..97e8418198c 100644
--- a/tmva/src/CrossEntropy.cxx
+++ b/tmva/src/CrossEntropy.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: CrossEntropy.cxx,v 1.10 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: CrossEntropy.cxx,v 1.11 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -14,7 +14,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/src/DataSet.cxx b/tmva/src/DataSet.cxx
index 2b437f287c8..ef65307cc67 100644
--- a/tmva/src/DataSet.cxx
+++ b/tmva/src/DataSet.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: DataSet.cxx,v 1.78 2006/11/14 23:02:57 stelzer Exp $
+// @(#)root/tmva $Id: DataSet.cxx,v 1.82 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,11 +13,11 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -56,14 +56,16 @@ TMVA::DataSet::DataSet()
fWeightFormula( 0 ),
fLogger( GetName(), kINFO )
{
- fDecorrMatrix[0] = fDecorrMatrix[1] = 0;
- fPrincipal[0] = fPrincipal[1] = 0;
+ // constructor
+
+ fDecorrMatrix[0] = fDecorrMatrix[1] = 0;
+ fPrincipal[0] = fPrincipal[1] = 0;
fPreprocessingEnabled[Types::kNone] = kTRUE;
fPreprocessingEnabled[Types::kDecorrelated] = fPreprocessingEnabled[Types::kPCA] = kFALSE;
for (Int_t corr=0; corr!=Types::kMaxPreprocessingMethod; corr++) fFlagPreprocessed[corr] = kFALSE;
- for (Int_t dim1=0; dim1!=kMaxTreeType; dim1++) {
+ for (Int_t dim1=0; dim1!=Types::kMaxTreeType; dim1++) {
for (Int_t dim2=0; dim2!=Types::kMaxSBType; dim2++) {
fDataStats[dim1][dim2]=0;
}
@@ -76,8 +78,11 @@ TMVA::DataSet::~DataSet()
}
//_______________________________________________________________________
-Bool_t TMVA::DataSet::ReadEvent(TTree* tr, UInt_t evidx, Types::PreprocessingMethod corr, Types::SBType type) const
+Bool_t TMVA::DataSet::ReadEvent(TTree* tr, UInt_t evidx, Types::EPreprocessingMethod corr, Types::ESBType type) const
{
+ // read event from a tree into memory
+ // after the reading the event transformation is called
+
if (tr == 0) fLogger << kFATAL << "<ReadEvent> zero Tree Pointer encountered" << Endl;
Bool_t needRead = kFALSE;
@@ -105,8 +110,9 @@ Bool_t TMVA::DataSet::ReadEvent(TTree* tr, UInt_t evidx, Types::PreprocessingMet
return ApplyTransformation( corr, (type == Types::kTrueType) ? Event().IsSignal() : (type == Types::kSignal) );
}
-Bool_t TMVA::DataSet::ApplyTransformation( Types::PreprocessingMethod corr, Bool_t useSignal ) const
+Bool_t TMVA::DataSet::ApplyTransformation( Types::EPreprocessingMethod corr, Bool_t useSignal ) const
{
+ // applies the transformation (none, decorellation, PCA) to the event data
switch (corr) {
@@ -148,6 +154,9 @@ Bool_t TMVA::DataSet::ApplyTransformation( Types::PreprocessingMethod corr, Bool
void TMVA::DataSet::ResetBranchAndEventAddresses( TTree* tree )
{
+ // resets all branch adresses of the tree given as parameter
+ // to the event memory
+
if (tree != 0) {
tree->ResetBranchAddresses();
if (fEvent != 0) fEvent->SetBranchAddresses( tree );
@@ -158,26 +167,37 @@ void TMVA::DataSet::ResetBranchAndEventAddresses( TTree* tree )
//_______________________________________________________________________
void TMVA::DataSet::AddSignalTree(TTree* tr, double weight)
{
+ // add a signal tree to the dataset to be used as input
+ // multiple trees is not used at the moment, use chains instead
fSignalTrees.push_back(TreeInfo(tr,weight));
}
//_______________________________________________________________________
void TMVA::DataSet::AddBackgroundTree(TTree* tr, double weight)
{
+ // add a background tree to the dataset to be used as input
+ // multiple trees is not used at the moment, use chains instead
fBackgroundTrees.push_back(TreeInfo(tr,weight));
}
//_______________________________________________________________________
void TMVA::DataSet::AddVariable(const TString& expression, char varType, void* external)
{
+ // add a variable (can be a complex expression) to the set of variables used in
+ // the MV analysis
fVariables.push_back(VariableInfo(expression, fVariables.size()+1, varType, external));
fVariableStrings.push_back(expression);
}
//_______________________________________________________________________
// corr==0 - correlated, corr==1 - decorrelated, corr==2 - PCAed
-void TMVA::DataSet::CalcNorm(Types::PreprocessingMethod corr)
+void TMVA::DataSet::CalcNorm(Types::EPreprocessingMethod corr)
{
+ // method to calculate minimum, maximum, mean, and RMS for all
+ // variables used in the MVA for the given data preprocessing
+ // method (none, decorrelate, PCA)
+
+ // preprocessing method has to be enabled
if(!PreprocessingEnabled(corr)) return;
if (GetTrainingTree()==0) return;
@@ -193,11 +213,11 @@ void TMVA::DataSet::CalcNorm(Types::PreprocessingMethod corr)
TVectorD x0( nvar ); x0 *= 0;
for (UInt_t ievt=0; ievt<nevts; ievt++) {
- ReadTrainingEvent(ievt,(Types::PreprocessingMethod)corr);
+ ReadTrainingEvent(ievt,(Types::EPreprocessingMethod)corr);
for (UInt_t ivar=0; ivar<nvar; ivar++) {
Double_t x = Event().GetVal(ivar);
- UpdateNorm( ivar, x, (Types::PreprocessingMethod)corr);
+ UpdateNorm( ivar, x, (Types::EPreprocessingMethod)corr);
x2(ivar) += x*x;
x0(ivar) += x;
}
@@ -221,6 +241,7 @@ void TMVA::DataSet::CalcNorm(Types::PreprocessingMethod corr)
//_______________________________________________________________________
Int_t TMVA::DataSet::FindVar(const TString& var) const
{
+ // find variable by name
for (UInt_t ivar=0; ivar<GetNVariables(); ivar++)
if (var == GetInternalVarName(ivar)) return ivar;
fLogger << kFATAL << "<FindVar> variable \'" << var << "\' not found" << Endl;
@@ -228,8 +249,9 @@ Int_t TMVA::DataSet::FindVar(const TString& var) const
}
//_______________________________________________________________________
-void TMVA::DataSet::UpdateNorm ( Int_t ivar, Double_t x, Types::PreprocessingMethod corr)
+void TMVA::DataSet::UpdateNorm ( Int_t ivar, Double_t x, Types::EPreprocessingMethod corr)
{
+ // update min and max of a given variable and a given preprocessing method
if (x < GetXmin( ivar, corr )) SetXmin( ivar, x, corr );
if (x > GetXmax( ivar, corr )) SetXmax( ivar, x, corr );
}
@@ -237,6 +259,17 @@ void TMVA::DataSet::UpdateNorm ( Int_t ivar, Double_t x, Types::PreprocessingMe
//_______________________________________________________________________
void TMVA::DataSet::PrepareForTrainingAndTesting( Int_t Ntrain, Int_t Ntest, TString TreeName )
{
+ // The internally used training and testing trees are prepaired in
+ // this method
+ // First the variables (expressions) of interest are copied from
+ // the given signal and background trees/chains into the local
+ // trees (training and testing), according to the specified numbers
+ // of training and testing events
+ // Second DataSet::CalcNorm is called to determine min, max, mean,
+ // and rms for all variables
+ // Optionally (if specified as option) the decorrelation and PCA
+ // preparation is executed
+
fLogger << kINFO << "prepare training and Test samples" << Endl;
fLogger << kINFO << "" << NSignalTrees()
<< " signal trees with total number of events : " << flush;
@@ -275,7 +308,9 @@ void TMVA::DataSet::PrepareForTrainingAndTesting( Int_t Ntrain, Int_t Ntest, TSt
if (Ntrain > 0 && Ntest == 0) {
array[0] = Ntrain;
- nsig_train = TMath::MinElement(3,array);
+ // nsig_train = TMath::MinElement(3,array);
+ // to be backward compatible with ROOT 4.02
+ nsig_train = TMath::Min(array[0],TMath::Min(array[1],array[2]));
nbkg_train = nsig_train;
nsig_test_min = nsig_train;
nbkg_test_min = nsig_train;
@@ -291,7 +326,9 @@ void TMVA::DataSet::PrepareForTrainingAndTesting( Int_t Ntrain, Int_t Ntest, TSt
}
else if (Ntrain > 0 && Ntest > 0) {
array[0] = Ntrain;
- nsig_train = TMath::MinElement(3,array);
+ // nsig_train = TMath::MinElement(3,array);
+ // to be backward compatible with ROOT 4.02
+ nsig_train = TMath::Min(array[0],TMath::Min(array[1],array[2]));
nbkg_train = nsig_train;
nsig_test_min = nsig_train;
nbkg_test_min = nsig_train;
@@ -492,22 +529,22 @@ void TMVA::DataSet::PrepareForTrainingAndTesting( Int_t Ntrain, Int_t Ntest, TSt
ac++;
if ( ac <= n_train[sb]) {
trainingTree->Fill();
- fDataStats[kTraining][Types::kSBBoth]++;
- fDataStats[kTraining][sb]++;
+ fDataStats[Types::kTraining][Types::kSBBoth]++;
+ fDataStats[Types::kTraining][sb]++;
}
if ((ac > n_test_min[sb])&& (ac <= n_test[sb])) {
testTree->Fill();
- fDataStats[kTesting][Types::kSBBoth]++;
- fDataStats[kTesting][sb]++;
+ fDataStats[Types::kTesting][Types::kSBBoth]++;
+ fDataStats[Types::kTesting][sb]++;
}
}
}
tr[sb]->ResetBranchAddresses();
// some output
- fLogger << kINFO << "collected " << fDataStats[kTraining][sb] << " " << kindS[sb]
+ fLogger << kINFO << "collected " << fDataStats[Types::kTraining][sb] << " " << kindS[sb]
<< " events for the training sample" << Endl;
- fLogger << kINFO << "collected " << fDataStats[kTesting][sb] << " " << kindS[sb]
+ fLogger << kINFO << "collected " << fDataStats[Types::kTesting][sb] << " " << kindS[sb]
<< " events for the test sample" << Endl;
}
@@ -560,8 +597,16 @@ void TMVA::DataSet::PrepareForTrainingAndTesting( Int_t Ntrain, Int_t Ntest, TSt
}
//_______________________________________________________________________
-Bool_t TMVA::DataSet::Preprocess( Types::PreprocessingMethod corr )
+Bool_t TMVA::DataSet::Preprocess( Types::EPreprocessingMethod corr )
{
+ // the dataset can be preprocessed with one of three methods
+ // - None: trivial identity
+ // - Decorellation: transformation that decorellates the input variables
+ // - PCA: Principal component analysis
+ // first the transformations are calculated, then the minimum,
+ // maximum, mean, and rms for the transformed set of variables are
+ // calculated (in DataSet::CalcNorm)
+
if( corr == Types::kNone ) return kTRUE;
if( ! PreprocessingEnabled(corr) ) return kFALSE;
@@ -608,6 +653,12 @@ Bool_t TMVA::DataSet::Preprocess( Types::PreprocessingMethod corr )
//_______________________________________________________________________
void TMVA::DataSet::ChangeToNewTree( TTree* tr )
{
+ // While the data gets copied into the local training and testing
+ // trees, the input tree can change (for intance when changing from
+ // signal to background tree, or using TChains as input) The
+ // TTreeFormulas, that hold the input expressions need to be
+ // reassociated with the new tree, which is done here
+
vector<TTreeFormula*>::const_iterator varFIt = fInputVarFormulas.begin();
for (;varFIt!=fInputVarFormulas.end();varFIt++) delete *varFIt;
fInputVarFormulas.clear();
@@ -637,8 +688,11 @@ void TMVA::DataSet::ChangeToNewTree( TTree* tr )
}
//_______________________________________________________________________
-void TMVA::DataSet::PlotVariables( TString tree, TString folderName, Types::PreprocessingMethod corr )
+void TMVA::DataSet::PlotVariables( TString tree, TString folderName, Types::EPreprocessingMethod corr )
{
+ // wrapper around PlotVariable(TTree*) that allows to call
+ // PlotVariables with a name
+
if(!PreprocessingEnabled(corr)) return;
tree.ToLower();
if (tree.BeginsWith("train")) {
@@ -650,8 +704,12 @@ void TMVA::DataSet::PlotVariables( TString tree, TString folderName, Types::Prep
}
//_______________________________________________________________________
-void TMVA::DataSet::PlotVariables( TTree* theTree, TString folderName, Types::PreprocessingMethod corr )
+void TMVA::DataSet::PlotVariables( TTree* theTree, TString folderName, Types::EPreprocessingMethod corr )
{
+ // create histograms from the input variables
+ // - histograms for all input variables
+ // - scatter plots for all pairs of input variables
+
if(!PreprocessingEnabled(corr)) return;
// if decorrelation has not been achieved, the decorrelation tree may be empty
@@ -682,7 +740,7 @@ void TMVA::DataSet::PlotVariables( TTree* theTree, TString folderName, Types::Pr
TVectorD rmsS( nvar ), meanS( nvar );
TVectorD rmsB( nvar ), meanB( nvar );
- UInt_t nevts = theTree->GetEntries();
+ UInt_t nevts = (UInt_t)theTree->GetEntries();
UInt_t nS = 0, nB = 0;
for (UInt_t ievt=0; ievt<nevts; ievt++) {
ReadTrainingEvent( ievt, corr, Types::kTrueType );
@@ -996,8 +1054,12 @@ Double_t TMVA::DataSet::GetSeparation( TH1* S, TH1* B ) const
}
//_______________________________________________________________________
-Bool_t TMVA::DataSet::PreparePreprocessing( Types::PreprocessingMethod corr, TTree* originalTree )
+Bool_t TMVA::DataSet::PreparePreprocessing( Types::EPreprocessingMethod corr, TTree* originalTree )
{
+ // For the preprocessing methods kDecorrelate and kPCA the
+ // transformation matrices are calculated. For the PCA the class
+ // TPrincipal is used.
+
if( originalTree == 0 ) return kFALSE;
// creates a deep copy of a tree with all of the values decorrelated
@@ -1179,51 +1241,57 @@ void TMVA::DataSet::GetSQRMats( TMatrixD*& sqS, TMatrixD*& sqB, vector<TString>*
//_______________________________________________________________________
void TMVA::DataSet::CalculatePrincipalComponents (TTree* originalTree, TPrincipal *&sigPrincipal, TPrincipal *&bgdPrincipal, vector<TString>* theVars )
{
- if (sigPrincipal != NULL) { delete sigPrincipal; sigPrincipal = 0; }
- if (bgdPrincipal != NULL) { delete bgdPrincipal; sigPrincipal = 0; }
-
- Int_t nvar = (int)theVars->size();
- sigPrincipal = new TPrincipal (nvar, ""); // Not normalizing and not storing input data, for performance reasons. Should perhaps restore normalization.
- bgdPrincipal = new TPrincipal (nvar, "");
-
- // Should we shove this into TMVA::Tools?
-
- TObjArrayIter branchIter( originalTree->GetListOfBranches(), kIterForward );
- TBranch* branch = NULL;
- Long64_t ievt, entries = originalTree->GetEntries();
- Float_t * fvec = new Float_t [nvar];
- Double_t * dvec = new Double_t [nvar];
- Int_t type, jvar=-1;
- Float_t weight, boostweight;
-
- while ((branch = (TBranch*)branchIter.Next()) != 0) {
- if ((TString)branch->GetName() == "type")
- originalTree->SetBranchAddress( branch->GetName(), &type );
- else if ((TString)branch->GetName() == "weight")
- originalTree->SetBranchAddress( branch->GetName(), &weight );
- else if ((TString)branch->GetName() == "boostweight")
- originalTree->SetBranchAddress( branch->GetName(), &boostweight );
- else
- originalTree->SetBranchAddress( branch->GetName(), &fvec[++jvar] );
- }
-
- for (ievt=0; ievt<entries; ievt++) {
- originalTree->GetEntry( ievt );
- TPrincipal *princ = type == Types::kSignal ? sigPrincipal : bgdPrincipal;
- for (Int_t i = 0; i < nvar; i++)
- dvec [i] = (Double_t) fvec [i];
- princ->AddRow (dvec);
- }
-
- sigPrincipal->MakePrincipals();
- bgdPrincipal->MakePrincipals();
-
- delete fvec; delete dvec;
+ // calculate the principal components for the signal and the background data
+ // it uses the MakePrincipal method of ROOT's TPrincipal class
+
+ if (sigPrincipal != NULL) { delete sigPrincipal; sigPrincipal = 0; }
+ if (bgdPrincipal != NULL) { delete bgdPrincipal; sigPrincipal = 0; }
+
+ Int_t nvar = (int)theVars->size();
+ sigPrincipal = new TPrincipal (nvar, ""); // Not normalizing and not storing input data, for performance reasons. Should perhaps restore normalization.
+ bgdPrincipal = new TPrincipal (nvar, "");
+
+ // Should we shove this into TMVA::Tools?
+
+ TObjArrayIter branchIter( originalTree->GetListOfBranches(), kIterForward );
+ TBranch* branch = NULL;
+ Long64_t ievt, entries = originalTree->GetEntries();
+ Float_t * fvec = new Float_t [nvar];
+ Double_t * dvec = new Double_t [nvar];
+ Int_t type, jvar=-1;
+ Float_t weight, boostweight;
+
+ while ((branch = (TBranch*)branchIter.Next()) != 0) {
+ if ((TString)branch->GetName() == "type")
+ originalTree->SetBranchAddress( branch->GetName(), &type );
+ else if ((TString)branch->GetName() == "weight")
+ originalTree->SetBranchAddress( branch->GetName(), &weight );
+ else if ((TString)branch->GetName() == "boostweight")
+ originalTree->SetBranchAddress( branch->GetName(), &boostweight );
+ else
+ originalTree->SetBranchAddress( branch->GetName(), &fvec[++jvar] );
+ }
+
+ for (ievt=0; ievt<entries; ievt++) {
+ originalTree->GetEntry( ievt );
+ TPrincipal *princ = type == Types::kSignal ? sigPrincipal : bgdPrincipal;
+ for (Int_t i = 0; i < nvar; i++)
+ dvec [i] = (Double_t) fvec [i];
+ princ->AddRow (dvec);
+ }
+
+ sigPrincipal->MakePrincipals();
+ bgdPrincipal->MakePrincipals();
+
+ delete fvec; delete dvec;
}
//_______________________________________________________________________
-void TMVA::DataSet::WriteVarsToStream(std::ostream& o, Types::PreprocessingMethod corr) const
+void TMVA::DataSet::WriteVarsToStream(std::ostream& o, Types::EPreprocessingMethod corr) const
{
+ // write the list of variables (name, min, max) for a given data
+ // transformation method to the stream
+
o << "NVar " << GetNVariables() << endl;
std::vector<VariableInfo>::const_iterator varIt = fVariables.begin();
for (;varIt!=fVariables.end(); varIt++) varIt->WriteToStream(o,corr);
@@ -1232,6 +1300,7 @@ void TMVA::DataSet::WriteVarsToStream(std::ostream& o, Types::PreprocessingMetho
//_______________________________________________________________________
void TMVA::DataSet::WriteCorrMatToStream(std::ostream& o) const
{
+ // write the decorrelation matrix to the stream
for (Int_t matType=0; matType<2; matType++) {
o << "# correlation matrix " << endl;
TMatrixD* mat = fDecorrMatrix[matType];
@@ -1247,8 +1316,11 @@ void TMVA::DataSet::WriteCorrMatToStream(std::ostream& o) const
}
//_______________________________________________________________________
-void TMVA::DataSet::ReadVarsFromStream(std::istream& istr, Types::PreprocessingMethod corr)
+void TMVA::DataSet::ReadVarsFromStream(std::istream& istr, Types::EPreprocessingMethod corr)
{
+ // Read the variables (name, min, max) for a given data
+ // transformation method from the stream.
+
TString dummy;
Int_t readNVar;
istr >> dummy >> readNVar;
@@ -1293,6 +1365,8 @@ void TMVA::DataSet::ReadVarsFromStream(std::istream& istr, Types::PreprocessingM
//_______________________________________________________________________
void TMVA::DataSet::ReadCorrMatFromStream(std::istream& istr )
{
+ // Read the decorellation matrix from an input stream
+
char buf[512];
istr.getline(buf,512);
TString strvar, dummy;
diff --git a/tmva/src/DecisionTree.cxx b/tmva/src/DecisionTree.cxx
index 4eba30e6cfb..9892652601a 100644
--- a/tmva/src/DecisionTree.cxx
+++ b/tmva/src/DecisionTree.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: DecisionTree.cxx,v 1.45 2006/11/14 15:21:00 stelzer Exp $
+// @(#)root/tmva $Id: DecisionTree.cxx,v 1.48 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -27,7 +27,7 @@
* (http://mva.sourceforge.net/license.txt) *
* *
* File and Version Information: *
- * $Id: DecisionTree.cxx,v 1.45 2006/11/14 15:21:00 stelzer Exp $
+ * $Id: DecisionTree.cxx,v 1.48 2006/11/16 22:51:58 helgevoss Exp $
**********************************************************************************/
//_______________________________________________________________________
@@ -172,7 +172,8 @@ TMVA::DecisionTree::~DecisionTree( void )
}
//_______________________________________________________________________
-void TMVA::DecisionTree::SetParentTreeInNodes( DecisionTreeNode *n){
+void TMVA::DecisionTree::SetParentTreeInNodes( DecisionTreeNode *n)
+{
// descend a tree to find all its leaf nodes, fill max depth reached in the
// tree at the same time.
@@ -354,7 +355,8 @@ void TMVA::DecisionTree::FillEvent( TMVA::Event & event,
//_______________________________________________________________________
-void TMVA::DecisionTree::ClearTree(){
+void TMVA::DecisionTree::ClearTree()
+{
// clear the tree nodes (their S/N, Nevents etc), just keep the structure of the tree
if (this->GetRoot()!=NULL)
@@ -364,7 +366,8 @@ void TMVA::DecisionTree::ClearTree(){
//_______________________________________________________________________
-void TMVA::DecisionTree::PruneTree(){
+void TMVA::DecisionTree::PruneTree()
+{
// prune (get rid of internal nodes) the Decision tree to avoid overtraining
// serveral different pruning methods can be applied as selected by the
// variable "fPruneMethod". Currently however only the Expected Error Pruning
@@ -387,7 +390,8 @@ void TMVA::DecisionTree::PruneTree(){
//_______________________________________________________________________
-void TMVA::DecisionTree::PruneTreeEEP(DecisionTreeNode *node){
+void TMVA::DecisionTree::PruneTreeEEP(DecisionTreeNode *node)
+{
// recursive prunig of nodes using the Expected Error Pruning (EEP)
// if internal node, then prune
DecisionTreeNode *l = (DecisionTreeNode*)node->GetLeft();
@@ -404,7 +408,8 @@ void TMVA::DecisionTree::PruneTreeEEP(DecisionTreeNode *node){
}
//_______________________________________________________________________
-void TMVA::DecisionTree::PruneTreeCC(){
+void TMVA::DecisionTree::PruneTreeCC()
+{
// prunig of nodes using the Cost Complexity criteria. The Pruning is performed
// until a minimum in the cost complexity CC(alpha) is reached.
// CC(alpha) = alpha*NLeafs + sum_over_leafs[ N*Quality(leaf) ]
@@ -429,7 +434,8 @@ void TMVA::DecisionTree::PruneTreeCC(){
//_______________________________________________________________________
-void TMVA::DecisionTree::PruneTreeMCC(){
+void TMVA::DecisionTree::PruneTreeMCC()
+{
// Similar to the CostCoplexity pruning, only here I calculate immediately
// the "prunestrength" (= alpha, the regularisation parameter in the CostComplexity)
// for which the respective subtree below a node would be pruned. Then I continue
@@ -768,39 +774,39 @@ Double_t TMVA::DecisionTree::GetNodeError(DecisionTreeNode *node)
// this node) is (1-f)
// now f has a statistical error according to the binomial distribution
// hence the error on f can be estimated (same error as the binomial error
- // for efficency calculations ( sigma = sqrt(eff(1-eff)/N) )
+ // for efficency calculations ( sigma = sqrt(eff(1-eff)/nEvts ) )
- Double_t errorRate=0;
+ Double_t errorRate = 0;
- Double_t N=node->GetNEvents();
+ Double_t nEvts = node->GetNEvents();
//fraction of correctly classified events by this node:
Double_t f=0;
if (node->GetSoverSB() > 0.5) f = node->GetSoverSB();
else f = (1-node->GetSoverSB());
- Double_t df = sqrt(f*(1-f)/N);
+ Double_t df = sqrt(f*(1-f)/nEvts );
errorRate = std::min(1.,(1 - (f-fPruneStrength*df) ));
// -------------------------------------------------------------------
// Minimum Error Pruning (MEP) accordig to Niblett/Bratko
//# of correctly classified events by this node:
- //Double_t n=f*N;
+ //Double_t n=f*nEvts ;
//Double_t p_apriori = 0.5, m=100;
- //errorRate = (N - n + (1-p_apriori) * m ) / (N + m);
+ //errorRate = (nEvts - n + (1-p_apriori) * m ) / (nEvts + m);
// Pessimistic error Pruing (proposed by Quinlan (error estimat with continuity approximation)
//# of correctly classified events by this node:
- //Double_t n=f*N;
- //errorRate = (N - n + 0.5) / N;
+ //Double_t n=f*nEvts ;
+ //errorRate = (nEvts - n + 0.5) / nEvts ;
//const Double Z=.65;
//# of correctly classified events by this node:
- //Double_t n=f*N;
- //errorRate = (f + Z*Z/(2*N) + Z*sqrt(f/N - f*f/N + Z*Z/4/N/N) ) / (1 + Z*Z/N);
- //errorRate = (n + Z*Z/2 + Z*sqrt(n - n*n/N + Z*Z/4) )/ (N + Z*Z);
+ //Double_t n=f*nEvts ;
+ //errorRate = (f + Z*Z/(2*nEvts ) + Z*sqrt(f/nEvts - f*f/nEvts + Z*Z/4/nEvts /nEvts ) ) / (1 + Z*Z/nEvts );
+ //errorRate = (n + Z*Z/2 + Z*sqrt(n - n*n/nEvts + Z*Z/4) )/ (nEvts + Z*Z);
//errorRate = 1 - errorRate;
// -------------------------------------------------------------------
diff --git a/tmva/src/DecisionTreeNode.cxx b/tmva/src/DecisionTreeNode.cxx
index da5c667d7ca..aafd2a591a7 100644
--- a/tmva/src/DecisionTreeNode.cxx
+++ b/tmva/src/DecisionTreeNode.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: DecisionTreeNode.cxx,v 1.23 2006/11/14 15:03:46 stelzer Exp $
+// @(#)root/tmva $Id: DecisionTreeNode.cxx,v 1.25 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* CopyRight (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -144,7 +144,7 @@ Bool_t TMVA::DecisionTreeNode::GoesLeft(const TMVA::Event & e) const
Double_t TMVA::DecisionTreeNode::GetSoverSB( void ) const
{
// return the S/(S+B) for the node
- return this->GetNSigEvents() / ( this->GetNSigEvents() + this->GetNBkgEvents());
+ return this->GetNSigEvents() / ( this->GetNSigEvents() + this->GetNBkgEvents());
}
//_______________________________________________________________________
diff --git a/tmva/src/Event.cxx b/tmva/src/Event.cxx
index e0e69ecbd6c..cc125e18b58 100644
--- a/tmva/src/Event.cxx
+++ b/tmva/src/Event.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Event.cxx,v 1.26 2006/11/13 20:02:23 helgevoss Exp $
+// @(#)root/tmva $Id: Event.cxx,v 1.29 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -86,6 +86,9 @@ TMVA::Event::Event(const Event& event)
void TMVA::Event::InitPointers(bool AllowExternalLink)
{
+ // sets the links of fVarPtr to the internal arrays that hold the
+ // integer and float variables
+
fVarPtrI = new Int_t[fCountI];
fVarPtrF = new Float_t[fCountF];
@@ -97,14 +100,14 @@ void TMVA::Event::InitPointers(bool AllowExternalLink)
// set the void pointer (which are used to access the data) to the proper field
// if external field is given
if (AllowExternalLink&& var.GetExternalLink()!=0) {
- fVarPtr[ivar] = var.GetExternalLink();
- // or if its type is I(int) or F(float)
- }
- else if (var.VarType()=='F') {
+ fVarPtr[ivar] = var.GetExternalLink();
+ // or if its type is I(int) or F(float)
+ }
+ else if (var.VarType()=='F') {
// set the void pointer to the float field
fVarPtr[ivar] = fVarPtrF+ivarF++;
}
- else if (var.VarType()=='I') {
+ else if (var.VarType()=='I') {
// set the void pointer to the int field
fVarPtr[ivar] = fVarPtrI+ivarI++;
}
@@ -115,6 +118,9 @@ void TMVA::Event::InitPointers(bool AllowExternalLink)
//____________________________________________________________
void TMVA::Event::SetBranchAddresses(TTree *tr)
{
+ // sets the branch addresses of the associated
+ // tree to the local memory as given by fVarPtr
+
fBranches.clear();
Int_t ivar(0);
TBranch * br(0);
@@ -144,6 +150,8 @@ void TMVA::Event::CopyVarValues( const Event& other )
//____________________________________________________________
void TMVA::Event::SetVal(UInt_t ivar, Float_t val)
{
+ // set variable ivar to val
+
if (ivar>=GetNVars())
fLogger << kFATAL << "<SetVal> cannot set value for variable index " << ivar
<< ", exceeds max index " << GetNVars()-1 << Endl;
@@ -154,12 +162,16 @@ void TMVA::Event::SetVal(UInt_t ivar, Float_t val)
//____________________________________________________________
Float_t TMVA::Event::GetValueNormalized(Int_t ivar) const
{
+ // returns the value of variable ivar, normalized to [-1,1]
+
return Tools::NormVariable(GetVal(ivar),fVariables[ivar].GetMin(),fVariables[ivar].GetMax());
}
//____________________________________________________________
void TMVA::Event::Print(std::ostream& o) const
{
+ // print method
+
o << fVariables.size() << " vars: ";
for(UInt_t ivar=0; ivar<fVariables.size(); ivar++)
o << std::setw(10) << GetVal(ivar);
diff --git a/tmva/src/Factory.cxx b/tmva/src/Factory.cxx
index fbf6ad91d38..2f666ebe9ba 100644
--- a/tmva/src/Factory.cxx
+++ b/tmva/src/Factory.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Factory.cxx,v 1.80 2006/11/15 00:20:32 stelzer Exp $
+// @(#)root/tmva $Id: Factory.cxx,v 1.85 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -74,15 +74,16 @@ const int MinNoTrainingEvents = 10;
const int MinNoTestEvents = 1;
const long int basketsize = 1280000;
-const TString BCwhite__f ( "\033[1;37m" );
-const TString BCred__f ( "\033[31m" );
-const TString BCblue__f ( "\033[34m" );
-const TString BCblue__b ( "\033[44m" );
-const TString BCred__b ( "\033[1;41m" );
-const TString EC__ ( "\033[0m" );
-const TString BClblue__b ( "\033[1;44m" );
-const TString BC_yellow = "\033[1;33m";
-const TString BC_lgreen = "\033[1;32m";
+const TString BCwhite__f = "\033[1;37m";
+const TString BCred__f = "\033[31m";
+const TString BCblue__f = "\033[34m";
+const TString BCblue__b = "\033[44m";
+const TString BCred__b = "\033[1;41m";
+const TString EC__ = "\033[0m";
+const TString BClblue__b = "\033[1;44m";
+const TString BC_yellow = "\033[1;33m";
+const TString BC_lgreen = "\033[1;32m";
+const TString BC_green = "\033[32m";
using namespace std;
@@ -517,7 +518,7 @@ void TMVA::Factory::PrepareTrainingAndTestTree( TCut cut, Int_t Ntrain, Int_t Nt
// the samples.
// ------------------------------------------------------------------------
//
- fLogger << kINFO << BC_lgreen << "preparing trees for training and testing..." << EC__ << Endl;
+ fLogger << kINFO << "preparing trees for training and testing..." << Endl;
if(fMultipleMVAs) Data().SetMultiCut(cut);
else Data().SetCut(cut);
@@ -566,7 +567,7 @@ Bool_t TMVA::Factory::BookMethod( TString theMethodName, TString methodTitle, TS
}
//_______________________________________________________________________
-Bool_t TMVA::Factory::BookMethod( TMVA::Types::MVA theMethod, TString methodTitle, TString theOption )
+Bool_t TMVA::Factory::BookMethod( TMVA::Types::EMVA theMethod, TString methodTitle, TString theOption )
{
// books MVA method; the option configuration string is custom for each MVA
// the TString field "theNameAppendix" serves to define (and distringuish)
@@ -577,31 +578,31 @@ Bool_t TMVA::Factory::BookMethod( TMVA::Types::MVA theMethod, TString methodTitl
// initialize methods
switch(theMethod) {
- case TMVA::Types::Cuts:
+ case TMVA::Types::kCuts:
method = new TMVA::MethodCuts ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::Fisher:
+ case TMVA::Types::kFisher:
method = new TMVA::MethodFisher ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::MLP:
+ case TMVA::Types::kMLP:
method = new TMVA::MethodMLP ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::TMlpANN:
+ case TMVA::Types::kTMlpANN:
method = new TMVA::MethodTMlpANN ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::CFMlpANN:
+ case TMVA::Types::kCFMlpANN:
method = new TMVA::MethodCFMlpANN ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::Likelihood:
+ case TMVA::Types::kLikelihood:
method = new TMVA::MethodLikelihood ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::Variable:
+ case TMVA::Types::kVariable:
method = new TMVA::MethodVariable ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::HMatrix:
+ case TMVA::Types::kHMatrix:
method = new TMVA::MethodHMatrix ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::PDERS:
+ case TMVA::Types::kPDERS:
method = new TMVA::MethodPDERS ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::BDT:
+ case TMVA::Types::kBDT:
method = new TMVA::MethodBDT ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::SVM:
+ case TMVA::Types::kSVM:
method = new TMVA::MethodSVM ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::RuleFit:
+ case TMVA::Types::kRuleFit:
method = new TMVA::MethodRuleFit ( fJobName, methodTitle, Data(), theOption ); break;
- case TMVA::Types::BayesClassifier:
+ case TMVA::Types::kBayesClassifier:
method = new TMVA::MethodBayesClassifier( fJobName, methodTitle, Data(), theOption ); break;
default:
fLogger << kFATAL << "method: " << theMethod << " does not exist" << Endl;
@@ -613,8 +614,8 @@ Bool_t TMVA::Factory::BookMethod( TMVA::Types::MVA theMethod, TString methodTitl
}
//_______________________________________________________________________
-Bool_t TMVA::Factory::BookMethod( TMVA::Types::MVA theMethod, TString methodTitle, TString methodOption,
- TMVA::Types::MVA theCommittee, TString committeeOption )
+Bool_t TMVA::Factory::BookMethod( TMVA::Types::EMVA theMethod, TString methodTitle, TString methodOption,
+ TMVA::Types::EMVA theCommittee, TString committeeOption )
{
// books MVA method; the option configuration string is custom for each MVA
// the TString field "theNameAppendix" serves to define (and distringuish)
@@ -625,7 +626,7 @@ Bool_t TMVA::Factory::BookMethod( TMVA::Types::MVA theMethod, TString methodTitl
// initialize methods
switch(theMethod) {
- case TMVA::Types::Committee:
+ case TMVA::Types::kCommittee:
method = new TMVA::MethodCommittee( fJobName, methodTitle, Data(), methodOption, theCommittee, committeeOption ); break;
default:
fLogger << kFATAL << "method: " << theMethod << " does not exist" << Endl;
@@ -653,7 +654,7 @@ TMVA::IMethod* TMVA::Factory::GetMVA( TString method )
void TMVA::Factory::TrainAllMethods( void )
{
// iterates over all MVAs that have been booked, and calls their training methods
- fLogger << kINFO << BC_lgreen << "training all methods..." << EC__ << Endl;
+ fLogger << kINFO << "training all methods..." << Endl;
// if multiple MVAs
if (fMultipleMVAs && !fMultipleStoredOptions ) {
@@ -699,7 +700,7 @@ void TMVA::Factory::TrainAllMethods( void )
void TMVA::Factory::TestAllMethods( void )
{
// iterates over all MVAs that have been booked, and calls their testing methods
- fLogger << kINFO << BC_lgreen << "testing all methods..." << EC__ << Endl;
+ fLogger << kINFO << "testing all methods..." << Endl;
// if multiple MVAs
if (fMultipleMVAs && !fMultipleStoredOptions ) {
@@ -730,7 +731,7 @@ void TMVA::Factory::TestAllMethods( void )
void TMVA::Factory::EvaluateAllVariables( TString options )
{
// iterates over all MVA input varables and evaluates them
- fLogger << kINFO << BC_lgreen << "evaluating all variables..." << EC__ << Endl;
+ fLogger << kINFO << "evaluating all variables..." << Endl;
// if multiple MVAs
if (fMultipleMVAs && !fMultipleStoredOptions ) {
@@ -757,9 +758,10 @@ void TMVA::Factory::EvaluateAllVariables( TString options )
//_______________________________________________________________________
void TMVA::Factory::EvaluateAllMethods( void )
{
- fLogger << kINFO << BC_lgreen << "evaluating all methods..." << EC__ << Endl;
// iterates over all MVAs that have been booked, and calls their evaluation methods
+ fLogger << kINFO << "evaluating all methods..." << Endl;
+
// if multiple MVAs
if (fMultipleMVAs && !fMultipleStoredOptions ) {
fLogger << kINFO << "EvaluateAllMethods will be called for multiple MVAs " << Endl;
diff --git a/tmva/src/GeneticANN.cxx b/tmva/src/GeneticANN.cxx
index 8f90b14e3bc..00b1d7dce22 100644
--- a/tmva/src/GeneticANN.cxx
+++ b/tmva/src/GeneticANN.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticANN.cxx,v 1.3 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticANN.cxx,v 1.5 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Matt Jachowski, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Matt Jachowski <jachowski@stanford.edu> - Stanford University, USA *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -44,7 +44,7 @@ TMVA::GeneticANN::GeneticANN( Int_t size, std::vector<LowHigh_t*> ranges, TMVA::
{
// constructor
fMethodMLP = methodMLP;
-}
+}
//_______________________________________________________________________
Double_t TMVA::GeneticANN::FitnessFunction( const std::vector<Double_t>& parameters )
diff --git a/tmva/src/GeneticBase.cxx b/tmva/src/GeneticBase.cxx
index b2a1a92ed91..b713c800269 100644
--- a/tmva/src/GeneticBase.cxx
+++ b/tmva/src/GeneticBase.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticBase.cxx,v 1.17 2006/10/17 21:22:29 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticBase.cxx,v 1.19 2006/11/16 22:51:58 helgevoss Exp $
// Author: Peter Speckmayer
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -278,5 +278,7 @@ Bool_t TMVA::GeneticBase::HasConverged( Int_t steps, Double_t improvement )
//_______________________________________________________________________
void TMVA::GeneticBase::Finalize()
-{}
+{
+ // nothing so far...
+}
diff --git a/tmva/src/GeneticCuts.cxx b/tmva/src/GeneticCuts.cxx
index a6b1b579dee..a9d5852b4a5 100644
--- a/tmva/src/GeneticCuts.cxx
+++ b/tmva/src/GeneticCuts.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticCuts.cxx,v 1.15 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticCuts.cxx,v 1.17 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Matt Jachowski, Peter Speckmayer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -14,13 +14,13 @@
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Matt Jachowski <jachowski@stanford.edu> - Stanford University, USA *
* Peter Speckmayer <speckmay@mail.cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -47,7 +47,7 @@ TMVA::GeneticCuts::GeneticCuts( Int_t size, std::vector<LowHigh_t*> ranges,
{
// constructor
fMethodCuts = methodCuts;
-}
+}
//_______________________________________________________________________
Double_t TMVA::GeneticCuts::FitnessFunction( const std::vector<Double_t>& parameters )
diff --git a/tmva/src/GeneticGenes.cxx b/tmva/src/GeneticGenes.cxx
index d92d30af95f..6cff6dbd73d 100644
--- a/tmva/src/GeneticGenes.cxx
+++ b/tmva/src/GeneticGenes.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticGenes.cxx,v 1.12 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticGenes.cxx,v 1.13 2006/11/16 22:51:58 helgevoss Exp $
// Author: Peter Speckmayer
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/GeneticPopulation.cxx b/tmva/src/GeneticPopulation.cxx
index 7b559c1aa0d..dacf2ca7040 100644
--- a/tmva/src/GeneticPopulation.cxx
+++ b/tmva/src/GeneticPopulation.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticPopulation.cxx,v 1.17 2006/10/17 21:22:29 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticPopulation.cxx,v 1.18 2006/11/16 22:51:58 helgevoss Exp $
// Author: Peter Speckmayer
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/GeneticRange.cxx b/tmva/src/GeneticRange.cxx
index e4a06eb820e..2d4d9ba6dc4 100644
--- a/tmva/src/GeneticRange.cxx
+++ b/tmva/src/GeneticRange.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GeneticRange.cxx,v 1.11 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GeneticRange.cxx,v 1.12 2006/11/16 22:51:58 helgevoss Exp $
// Author: Peter Speckmayer
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -24,7 +24,7 @@
* (http://tmva.sourceforge.net/LICENSE) *
* *
* File and Version Information: *
- * $Id: GeneticRange.cxx,v 1.11 2006/10/10 17:43:51 andreas.hoecker Exp $
+ * $Id: GeneticRange.cxx,v 1.12 2006/11/16 22:51:58 helgevoss Exp $
**********************************************************************************/
//_______________________________________________________________________
diff --git a/tmva/src/GiniIndex.cxx b/tmva/src/GiniIndex.cxx
index e02d42bb19e..eda0ffe37f0 100644
--- a/tmva/src/GiniIndex.cxx
+++ b/tmva/src/GiniIndex.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: GiniIndex.cxx,v 1.10 2006/10/10 17:43:51 andreas.hoecker Exp $
+// @(#)root/tmva $Id: GiniIndex.cxx,v 1.11 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -22,7 +22,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/src/MethodANNBase.cxx b/tmva/src/MethodANNBase.cxx
index 52442f15c90..1a8240c4242 100644
--- a/tmva/src/MethodANNBase.cxx
+++ b/tmva/src/MethodANNBase.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodANNBase.cxx,v 1.47 2006/11/14 23:02:57 stelzer Exp $
+// @(#)root/tmva $Id: MethodANNBase.cxx,v 1.49 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Matt Jachowski
/**********************************************************************************
@@ -74,12 +74,12 @@ TMVA::MethodANNBase::MethodANNBase( TString jobName, TString methodTitle, DataSe
TString theOption, TDirectory* theTargetDir )
: TMVA::MethodBase( jobName, methodTitle, theData, theOption, theTargetDir )
{
- /* Note: Right now it is an option to choose the neuron input function,
- but only the input function "sum" leads to weight convergence --
- otherwise the weights go to nan and lead to an ABORT.
- */
-
// standard constructor
+ // Note: Right now it is an option to choose the neuron input function,
+ // but only the input function "sum" leads to weight convergence --
+ // otherwise the weights go to nan and lead to an ABORT.
+
+
InitANNBase();
DeclareOptions();
@@ -90,15 +90,26 @@ TMVA::MethodANNBase::MethodANNBase( DataSet & theData,
TString theWeightFile, TDirectory* theTargetDir )
: TMVA::MethodBase( theData, theWeightFile, theTargetDir )
{
- // weight file constructor
+ // construct the Method from the weight file
InitANNBase();
DeclareOptions();
}
+//______________________________________________________________________________
void TMVA::MethodANNBase::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // here the options valid for ALL MVA methods are declared.
+ // know options: NCycles=xx :the number of training cycles
+ // Normalize=kTRUE,kFALSe :if normalised in put variables should be used
+ // HiddenLayser="N-1,N-2" :the specification of the hidden layers
+ // NeuronType=sigmoid,tanh,radial,linar : the type of activation function
+ // used at the neuronn
+ //
+
+
DeclareOptionRef(fNcycles=3000,"NCycles","Number of training cycles");
DeclareOptionRef(fNormalize=kTRUE, "Normalize", "Normalize input variables");
DeclareOptionRef(fLayerSpec="N-1,N-2","HiddenLayers","Specification of the hidden layers");
@@ -121,8 +132,11 @@ void TMVA::MethodANNBase::DeclareOptions()
delete names;
}
+//______________________________________________________________________________
void TMVA::MethodANNBase::ProcessOptions()
{
+ // decode the options in the option string
+
MethodBase::ProcessOptions();
vector<Int_t>* layout = ParseLayoutString(fLayerSpec);
@@ -585,7 +599,7 @@ const TMVA::Ranking* TMVA::MethodANNBase::CreateRanking()
// figure out average value of variable i
Double_t meanS, meanB, rmsS, rmsB, xmin, xmax;
- Statistics( TMVA::Types::kTrain, varName,
+ Statistics( TMVA::Types::kTraining, varName,
meanS, meanB, rmsS, rmsB, xmin, xmax );
avgVal = (meanS + meanB) / 2.0; // change this into a real weighted average
diff --git a/tmva/src/MethodBDT.cxx b/tmva/src/MethodBDT.cxx
index 03496cdafdf..6a79f86e218 100644
--- a/tmva/src/MethodBDT.cxx
+++ b/tmva/src/MethodBDT.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodBDT.cxx,v 1.58 2006/11/14 23:02:57 stelzer Exp $
+// @(#)root/tmva $Id: MethodBDT.cxx,v 1.64 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -113,25 +113,27 @@ TMVA::MethodBDT::MethodBDT( TString jobName, TString methodTitle, DataSet& theDa
// the standard constructor for the "boosted decision trees"
//
// MethodBDT (Boosted Decision Trees) options:
- // nTrees: number of trees in the forest to be created
- // BoostType: the boosting type for the trees in the forest (AdaBoost e.t.c..)
+ // know options:
+ // nTrees=Int_t: number of trees in the forest to be created
+ // BoostType= the boosting type for the trees in the forest (AdaBoost e.t.c..)
+ // known: AdaBoost
+ // Bagging
// SeparationType the separation criterion applied in the node splitting
+ // known: GiniIndex
+ // MisClassificationError
+ // CrossEntropy
+ // SDivSqrtSPlusB
// nEventsMin: the minimum number of events in a node (leaf criteria, stop splitting)
- // nCuts: the number of steps in the optimisation of the cut for a node
+ // nCuts: the number of steps in the optimisation of the cut for a node
// UseYesNoLeaf decide if the classification is done simply by the node type, or the S/B
// (from the training) in the leaf node
// UseWeightedTrees use average classification from the trees, or have the individual trees
// trees in the forest weighted (e.g. log(boostweight) from AdaBoost
- // PruneMethod The Pruning method: Expected Error or Cost Complexity");
+ // PruneMethod The Pruning method:
+ // known: ExpectedError
+ // CostComplexity
+ // CostComplexity2
// PruneStrength a parameter to adjust the amount of pruning. Should be large enouth such that overtraining is avoided");
- //
- // known SeparationTypes are:
- // - MisClassificationError
- // - GiniIndex
- // - CrossEntropy
- // known BoostTypes are:
- // - AdaBoost
- // - Bagging
InitBDT(); // sets default values
DeclareOptions();
@@ -153,13 +155,17 @@ TMVA::MethodBDT::MethodBDT( TString jobName, TString methodTitle, DataSet& theDa
//book monitoring histograms (currently for AdaBost, only)
fBoostWeightHist = new TH1F("fBoostWeight","Ada Boost weights",100,1,100);
- fErrFractHist = new TH2F("fErrFractHist","error fraction vs tree number",
- fNTrees,0,fNTrees,50,0,0.5);
+ fBoostWeightVsTree = new TH1F("fBoostWeightVsTree","Ada Boost weights",fNTrees,0,fNTrees);
+
+ fErrFractHist = new TH1F("fErrFractHist","error fraction vs tree number",fNTrees,0,fNTrees);
+
+ fNodesBeforePruningVsTree = new TH1I("fNodesBeforePruning","nodes before pruning",fNTrees,0,fNTrees);
+ fNodesAfterPruningVsTree = new TH1I("fNodesAfterPruning","nodes after pruning",fNTrees,0,fNTrees);
+
fMonitorNtuple= new TTree("fMonitorNtuple","BDT variables");
fMonitorNtuple->Branch("iTree",&fITree,"iTree/I");
fMonitorNtuple->Branch("boostWeight",&fBoostWeight,"boostWeight/D");
fMonitorNtuple->Branch("errorFraction",&fErrorFraction,"errorFraction/D");
- fMonitorNtuple->Branch("nNodes",&fNnodes,"nNodes/I");
}
//_______________________________________________________________________
@@ -180,6 +186,29 @@ TMVA::MethodBDT::MethodBDT( DataSet& theData,
//_______________________________________________________________________
void TMVA::MethodBDT::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options:
+ // nTrees=Int_t: number of trees in the forest to be created
+ // BoostType= the boosting type for the trees in the forest (AdaBoost e.t.c..)
+ // known: AdaBoost
+ // Bagging
+ // SeparationType the separation criterion applied in the node splitting
+ // known: GiniIndex
+ // MisClassificationError
+ // CrossEntropy
+ // SDivSqrtSPlusB
+ // nEventsMin: the minimum number of events in a node (leaf criteria, stop splitting)
+ // nCuts: the number of steps in the optimisation of the cut for a node
+ // UseYesNoLeaf decide if the classification is done simply by the node type, or the S/B
+ // (from the training) in the leaf node
+ // UseWeightedTrees use average classification from the trees, or have the individual trees
+ // trees in the forest weighted (e.g. log(boostweight) from AdaBoost
+ // PruneMethod The Pruning method:
+ // known: ExpectedError
+ // CostComplexity
+ // CostComplexity2
+ // PruneStrength a parameter to adjust the amount of pruning. Should be large enouth such that overtraining is avoided");
+
DeclareOptionRef(fNTrees, "NTrees", "number of trees in the forest");
DeclareOptionRef(fBoostType, "BoostType", "boosting type for the trees in the forest");
AddPreDefVal(TString("AdaBoost"));
@@ -203,6 +232,8 @@ void TMVA::MethodBDT::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodBDT::ProcessOptions()
{
+ // the option string is decoded, for available options see "DeclareOptions"
+
MethodBase::ProcessOptions();
fSepTypeS.ToLower();
@@ -234,12 +265,12 @@ void TMVA::MethodBDT::InitBDT( void )
{
// common initialisation with defaults for the BDT-Method
SetMethodName( "BDT" );
- SetMethodType( TMVA::Types::BDT );
+ SetMethodType( TMVA::Types::kBDT );
SetTestvarName();
fNTrees = 200;
fBoostType = "AdaBoost";
- fNodeMinEvents = 5;
+ fNodeMinEvents = 10;
fNCuts = 20;
fPruneMethod = TMVA::DecisionTree::kMCC;
fPruneStrength = 5; // means automatic determination of the prune strength using a validation sample
@@ -294,6 +325,9 @@ void TMVA::MethodBDT::Train( void )
fLogger << kINFO << "will train "<< fNTrees << " Decision Trees ... patience please" << Endl;
TMVA::Timer timer( fNTrees, GetName() );
+ Int_t nNodesBeforePruningCount = 0;
+ Int_t nNodesAfterPruningCount = 0;
+
Int_t nNodesBeforePruning = 0;
Int_t nNodesAfterPruning = 0;
@@ -312,12 +346,12 @@ void TMVA::MethodBDT::Train( void )
//
// std::vector<Event*> sample;
// for (std::vector<Event*>::iterator iev=fEventSample.begin();
- // iev != fEventSample.end(); iev++){
- // if ((*iev)->GetWeight() > 0.1) sample.push_back(*iev);
+ // iev != fEventSample.end(); iev++){
+ // if ((*iev)->GetWeight() > 0.1) sample.push_back(*iev);
// }
- // fNnodes = fForest.back()->BuildTree(sample);
+ // nNodesBeforePruning = fForest.back()->BuildTree(sample);
- fNnodes = fForest.back()->BuildTree(fEventSample);
+ nNodesBeforePruning = fForest.back()->BuildTree(fEventSample);
if (itree==1 && fgDebugLevel==1){
//plot Cost Complexity versus #Nodes for increasing pruning strengths
@@ -377,7 +411,8 @@ void TMVA::MethodBDT::Train( void )
}
- nNodesBeforePruning +=fNnodes;
+ nNodesBeforePruningCount +=nNodesBeforePruning;
+ fNodesBeforePruningVsTree->SetBinContent(itree+1,nNodesBeforePruning);
fBoostWeights.push_back( this->Boost(fEventSample, fForest.back(), itree) );
fITree = itree;
@@ -403,15 +438,16 @@ void TMVA::MethodBDT::Train( void )
fForest[itree]->SetPruneStrength(fPruneStrength);
fForest[itree]->PruneTree();
}
- fNnodes = fForest[itree]->GetNNodes();
- nNodesAfterPruning +=fNnodes;
+ nNodesAfterPruning = fForest[itree]->GetNNodes();
+ nNodesAfterPruningCount += nNodesAfterPruning;
+ fNodesAfterPruningVsTree->SetBinContent(itree+1,nNodesAfterPruning);
alpha->SetBinContent(itree+1,fPruneStrength);
}
alpha->Write();
fLogger << kINFO << "<Train> average number of nodes before/after pruning : "
- << nNodesBeforePruning/fNTrees << " / "
- << nNodesAfterPruning/fNTrees
+ << nNodesBeforePruningCount/fNTrees << " / "
+ << nNodesAfterPruningCount/fNTrees
<< Endl;
// get elapsed time
@@ -474,7 +510,7 @@ Double_t TMVA::MethodBDT::PruneTree( TMVA::DecisionTree *dt, Int_t itree)
<< " if that stitill didn't work, TRY IT BY HAND"
<< " currently Prunestrenght= " << alpha
<< " stepsize " << fDeltaPruneStrength << " " << Endl;
- troubleCount = 0; // try again
+ troubleCount = 0; // try again
fPruneStrength = 3; // if it was for the first time..
}else{
forceStop=kTRUE;
@@ -626,7 +662,10 @@ Double_t TMVA::MethodBDT::AdaBoost( vector<TMVA::Event*> eventSample, TMVA::Deci
}
fBoostWeightHist->Fill(boostWeight);
- fErrFractHist->Fill(fForest.size(),err);
+
+ fBoostWeightVsTree->SetBinContent(fForest.size(),boostWeight);
+
+ fErrFractHist->SetBinContent(fForest.size(),err);
fBoostWeight = boostWeight;
fErrorFraction = err;
@@ -738,7 +777,10 @@ void TMVA::MethodBDT::WriteMonitoringHistosToFile( void ) const
BaseDir()->cd();
fBoostWeightHist->Write();
+ fBoostWeightVsTree->Write();
fErrFractHist->Write();
+ fNodesBeforePruningVsTree->Write();
+ fNodesAfterPruningVsTree->Write();
fMonitorNtuple->Write();
// (*fForest.begin())->DrawTree("ExampleTree")->Write();
diff --git a/tmva/src/MethodBase.cxx b/tmva/src/MethodBase.cxx
index 5f228c1068f..31c521e6a68 100644
--- a/tmva/src/MethodBase.cxx
+++ b/tmva/src/MethodBase.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodBase.cxx,v 1.86 2006/11/15 00:20:32 stelzer Exp $
+// @(#)root/tmva $Id: MethodBase.cxx,v 1.92 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -27,7 +27,7 @@
* (http://tmva.sourceforge.net/LICENSE) *
* *
* File and Version Information: *
- * $Id: MethodBase.cxx,v 1.86 2006/11/15 00:20:32 stelzer Exp $
+ * $Id: MethodBase.cxx,v 1.92 2006/11/17 14:59:23 stelzer Exp $
**********************************************************************************/
//_______________________________________________________________________
@@ -385,6 +385,17 @@ void TMVA::MethodBase::Init()
//_______________________________________________________________________
void TMVA::MethodBase::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // here the options valid for ALL MVA methods are declared.
+ // know options: Preprocess=None,Decorrelated,PCA to use decorrelated variables
+ // instead of the original ones
+ // PreprocessType=Signal,Background which decorrelation matrix to use
+ // in the method. Only the Likelihood
+ // Method can make proper use of independent
+ // transformations of signal and background
+ // V for Verbose output (!V) for non verbos
+ // H for Help
+
DeclareOptionRef(fUseDecorr, "D", "use-decorrelated-variables flag (for backward compatibility)");
DeclareOptionRef(fPreprocessingString="None", "Preprocess", "Variable Decorrelation Method");
@@ -403,6 +414,8 @@ void TMVA::MethodBase::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodBase::ProcessOptions()
{
+ // the option string is decoded, for availabel options see "DeclareOptions"
+
if (fPreprocessingString == "None") fPreprocessingMethod = Types::kNone;
else if (fPreprocessingString == "Decorrelate" ) fPreprocessingMethod = Types::kDecorrelated;
else if (fPreprocessingString == "PCA" ) fPreprocessingMethod = Types::kPCA;
@@ -425,7 +438,7 @@ void TMVA::MethodBase::ProcessOptions()
if( GetPreprocessingMethod() == Types::kDecorrelated ) {
- Types::PreprocessingMethod c = Types::kDecorrelated;
+ Types::EPreprocessingMethod c = Types::kDecorrelated;
Data().EnablePreprocess(Types::kDecorrelated);
if( Data().Preprocess(Types::kDecorrelated) ) {
// local copy of the variable ranges
@@ -438,7 +451,7 @@ void TMVA::MethodBase::ProcessOptions()
}
if( GetPreprocessingMethod() == Types::kPCA ) {
- Types::PreprocessingMethod c = Types::kPCA;
+ Types::EPreprocessingMethod c = Types::kPCA;
Data().EnablePreprocess(Types::kPCA);
if( Data().Preprocess(Types::kPCA) ) {
// local copy of the variable ranges
@@ -467,6 +480,9 @@ void TMVA::MethodBase::TrainMethod()
//_______________________________________________________________________
void TMVA::MethodBase::WriteStateToStream(std::ostream& o) const
{
+ // general method used in writing the header of the weight files where
+ // the used variables, preprocessing type etc. is specified
+
o << "#GEN -*-*-*-*-*-*-*-*-*-*-*- general info -*-*-*-*-*-*-*-*-*-*-*-" << endl << endl;
o << "Method : " << GetMethodName() << endl;
o << "Creator: " << gSystem->GetUserInfo()->fUser << endl;
@@ -547,8 +563,11 @@ void TMVA::MethodBase::ReadStateFromFile()
}
}
+//_______________________________________________________________________
void TMVA::MethodBase::ReadStateFromStream( std::istream& fin )
{
+ // read the header from the weight files of the different MVA methods
+
char buf[512];
// first read the method name
@@ -595,7 +614,7 @@ void TMVA::MethodBase::ReadStateFromStream( std::istream& fin )
if (0 != fXmaxNorm[corr]) delete fXmaxNorm[corr];
fXminNorm[corr] = new Double_t[Data().GetNVariables()];
fXmaxNorm[corr] = new Double_t[Data().GetNVariables()];
- Types::PreprocessingMethod c = (Types::PreprocessingMethod) corr;
+ Types::EPreprocessingMethod c = (Types::EPreprocessingMethod) corr;
for(UInt_t ivar=0; ivar<Data().GetNVariables(); ivar++) {
SetXmin(ivar, Data().GetXmin(ivar, c), c);
SetXmax(ivar, Data().GetXmax(ivar, c), c);
@@ -614,6 +633,7 @@ void TMVA::MethodBase::ReadStateFromStream( std::istream& fin )
//_______________________________________________________________________
Double_t TMVA::MethodBase::GetEventValNormalized(Int_t ivar) const
{
+ // return the normalized event variable (normalized to interval [0,1]
return Tools::NormVariable( Data().Event().GetVal(ivar),
GetXmin(ivar, GetPreprocessingMethod()),
GetXmax(ivar, GetPreprocessingMethod()));
@@ -622,6 +642,9 @@ Double_t TMVA::MethodBase::GetEventValNormalized(Int_t ivar) const
//_______________________________________________________________________
TDirectory * TMVA::MethodBase::BaseDir( void ) const
{
+ // returns the ROOT directory where info/histograms etc of the
+ // corresponding MVA method are stored
+
if (fBaseDir != 0) return fBaseDir;
TDirectory* dir = 0;
@@ -961,6 +984,7 @@ Double_t TMVA::MethodBase::GetEfficiency( TString theString, TTree *theTree )
return 0.5*(effS + effS_); // the mean between bin above and bin below
}
+//_______________________________________________________________________
Double_t TMVA::MethodBase::GetTrainingEfficiency( TString theString)
{
// fill background efficiency (resp. rejection) versus signal efficiency plots
@@ -1018,7 +1042,7 @@ Double_t TMVA::MethodBase::GetTrainingEfficiency( TString theString)
ReadTrainingEvent(ievt);
TH1* theHist = (Data().Event().IsSignal() ? fTrainEffS : fTrainEffB);
-
+
Double_t theVal = this->GetMvaValue();
for (Int_t bin=1; bin<=fNbinsH; bin++)
@@ -1285,14 +1309,19 @@ Double_t TMVA::MethodBase::GetmuTransform( TTree *theTree )
return intS; // return average mu-transform for signal
}
-void TMVA::MethodBase::Statistics( TMVA::Types::TreeType treeType, const TString& theVarName,
+//_______________________________________________________________________
+void TMVA::MethodBase::Statistics( TMVA::Types::ETreeType treeType, const TString& theVarName,
Double_t& meanS, Double_t& meanB,
Double_t& rmsS, Double_t& rmsB,
Double_t& xmin, Double_t& xmax,
Bool_t norm )
{
- Long64_t entries = ( (treeType == TMVA::Types::kTest ) ? Data().GetNEvtTest() :
- (treeType == TMVA::Types::kTrain) ? Data().GetNEvtTrain() : -1 );
+ // calculates rms,mean, xmin, xmax of the event variable
+ // this can be either done for the variables as they are or for
+ // normalised variables (in the range of 0-1) if "norm" is set to kTRUE
+
+ Long64_t entries = ( (treeType == TMVA::Types::kTesting ) ? Data().GetNEvtTest() :
+ (treeType == TMVA::Types::kTraining) ? Data().GetNEvtTrain() : -1 );
// sanity check
if (entries <=0)
@@ -1312,8 +1341,10 @@ void TMVA::MethodBase::Statistics( TMVA::Types::TreeType treeType, const TString
// loop over all training events
for (Int_t i = 0; i < entries; i++) {
- if (treeType == TMVA::Types::kTest ) ReadTestEvent(i);
- else ReadTrainingEvent(i);
+ if (treeType == TMVA::Types::kTesting )
+ ReadTestEvent(i);
+ else
+ ReadTrainingEvent(i);
Double_t theVar = (norm) ? GetEventValNormalized(varIndex) : GetEventVal(varIndex);
@@ -1395,6 +1426,8 @@ Double_t TMVA::MethodBase::GetEffForRoot( Double_t theCut )
//______________________________________________________________________
void TMVA::MethodBase::PrintOptions() const
{
+ // prints out the options set in the options string and the defaults
+
fLogger << kINFO << "the following options are set:" << Endl;
TListIter optIt( & ListOfOptions() );
fLogger << kINFO << "by User:" << Endl;
@@ -1413,6 +1446,8 @@ void TMVA::MethodBase::PrintOptions() const
//______________________________________________________________________
void TMVA::MethodBase::WriteOptionsToStream(ostream& o) const
{
+ // write options to output stream (e.g. in writing the MVA weight files
+
TListIter optIt( & ListOfOptions() );
o << "# Set by User:" << endl;
while (OptionBase * opt = (OptionBase *) optIt()) if (opt->IsSet()) { opt->Print(o); o << endl; }
@@ -1425,6 +1460,8 @@ void TMVA::MethodBase::WriteOptionsToStream(ostream& o) const
//______________________________________________________________________
void TMVA::MethodBase::ReadOptionsFromStream(istream& istr)
{
+ // read option back from the weight file
+
fOptions = "";
char buf[512];
istr.getline(buf,512);
diff --git a/tmva/src/MethodBayesClassifier.cxx b/tmva/src/MethodBayesClassifier.cxx
index fecc2a97735..bf2f5ad89ac 100644
--- a/tmva/src/MethodBayesClassifier.cxx
+++ b/tmva/src/MethodBayesClassifier.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodBayesClassifier.cxx,v 1.2 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodBayesClassifier.cxx,v 1.5 2006/11/17 00:21:35 stelzer Exp $
// Author: Marcin ....
/**********************************************************************************
@@ -17,7 +17,7 @@
* University of Houston, *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -60,15 +60,20 @@ void TMVA::MethodBayesClassifier::InitBayesClassifier( void )
{
// default initialisation
SetMethodName( "BayesClassifier" );
- SetMethodType( TMVA::Types::BayesClassifier );
+ SetMethodType( TMVA::Types::kBayesClassifier );
SetTestvarName();
}
+//_______________________________________________________________________
void TMVA::MethodBayesClassifier::DeclareOptions()
-{}
+{
+ // define the options (their key words) that can be set in the option string
+}
+//_______________________________________________________________________
void TMVA::MethodBayesClassifier::ProcessOptions()
{
+ // the option string is decoded, for availabel options see "DeclareOptions"
MethodBase::ProcessOptions();
}
@@ -90,12 +95,14 @@ void TMVA::MethodBayesClassifier::Train( void )
//_______________________________________________________________________
void TMVA::MethodBayesClassifier::WriteWeightsToStream( ostream & o ) const
{
+ // write the weight from the training to a file (stream)
o << "whatever" << endl;
}
//_______________________________________________________________________
void TMVA::MethodBayesClassifier::ReadWeightsFromStream( istream & istr )
{
+ // read back the training results from a file (stream)
if (istr.eof());
}
diff --git a/tmva/src/MethodCFMlpANN.cxx b/tmva/src/MethodCFMlpANN.cxx
index 847f9754aa6..625f29bf6f8 100644
--- a/tmva/src/MethodCFMlpANN.cxx
+++ b/tmva/src/MethodCFMlpANN.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCFMlpANN.cxx,v 1.32 2006/11/14 23:02:57 stelzer Exp $
+// @(#)root/tmva $Id: MethodCFMlpANN.cxx,v 1.35 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -168,6 +168,10 @@ TMVA::MethodCFMlpANN::MethodCFMlpANN( DataSet & theData,
//_______________________________________________________________________
void TMVA::MethodCFMlpANN::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options: NCycles=xx :the number of training cycles
+ // HiddenLayser="N-1,N-2" :the specification of the hidden layers
+
DeclareOptionRef(fNcycles=3000,"NCycles","Number of training cycles");
DeclareOptionRef(fLayerSpec="N-1,N-2","HiddenLayers","Specification of the hidden layers");
}
@@ -175,6 +179,8 @@ void TMVA::MethodCFMlpANN::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodCFMlpANN::ProcessOptions()
{
+ // decode the options in the option string
+
MethodBase::ProcessOptions();
fNodes = new Int_t[100]; // input layer
@@ -210,7 +216,7 @@ void TMVA::MethodCFMlpANN::InitCFMlpANN( void )
{
// default initialisation called by all constructors
SetMethodName( "CFMlpANN" );
- SetMethodType( TMVA::Types::CFMlpANN );
+ SetMethodType( TMVA::Types::kCFMlpANN );
SetTestvarName();
// initialize all pointers
@@ -381,7 +387,7 @@ void TMVA::MethodCFMlpANN::WriteWeightsToStream( std::ostream & o ) const
//_______________________________________________________________________
void TMVA::MethodCFMlpANN::ReadWeightsFromStream( istream & istr )
{
-
+ // read back the weight from the training from file (stream)
TString var;
// read number of variables and classes
diff --git a/tmva/src/MethodCFMlpANN_Utils.cxx b/tmva/src/MethodCFMlpANN_Utils.cxx
index 9130aeb4eb8..222c3b1b99c 100644
--- a/tmva/src/MethodCFMlpANN_Utils.cxx
+++ b/tmva/src/MethodCFMlpANN_Utils.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCFMlpANN_Utils.cxx,v 1.14 2006/11/14 23:02:57 stelzer Exp $
+// @(#)root/tmva $Id: MethodCFMlpANN_Utils.cxx,v 1.15 2006/11/16 22:51:58 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -31,13 +31,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -45,7 +45,7 @@
* (http://tmva.sourceforge.net/LICENSE) *
* *
* File and Version Information: *
- * $Id: MethodCFMlpANN_Utils.cxx,v 1.14 2006/11/14 23:02:57 stelzer Exp $
+ * $Id: MethodCFMlpANN_Utils.cxx,v 1.15 2006/11/16 22:51:58 helgevoss Exp $
**********************************************************************************/
//_______________________________________________________________________
diff --git a/tmva/src/MethodCommittee.cxx b/tmva/src/MethodCommittee.cxx
index 821c06ba4ca..18b5df39b93 100644
--- a/tmva/src/MethodCommittee.cxx
+++ b/tmva/src/MethodCommittee.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCommittee.cxx,v 1.10 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodCommittee.cxx,v 1.14 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -63,12 +63,13 @@ ClassImp(TMVA::MethodCommittee)
//_______________________________________________________________________
TMVA::MethodCommittee::MethodCommittee( TString jobName, TString committeeTitle, DataSet& theData,
TString committeeOptions,
- Types::MVA method, TString methodOptions,
+ Types::EMVA method, TString methodOptions,
TDirectory* theTargetDir )
: TMVA::MethodBase( jobName, committeeTitle, theData, committeeOptions, theTargetDir ),
fMemberType( method ),
fMemberOption( methodOptions )
{
+ // constructor
InitCommittee(); // sets default values
DeclareOptions();
@@ -105,6 +106,16 @@ TMVA::MethodCommittee::MethodCommittee( DataSet& theData,
//_______________________________________________________________________
void TMVA::MethodCommittee::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options:
+ // NMembers <string> number of members in the committee
+ // UseMemberDecision <bool> use signal information from event (otherwise assume signal)
+ // UseWeightedMembers <bool> use weighted trees or simple average in classification from the forest
+ //
+ // BoostType <string> boosting type
+ // available values are: AdaBoost <default>
+ // Bagging
+
DeclareOptionRef(fNMembers, "NMembers", "number of members in the committee");
DeclareOptionRef(fUseMemberDecision=kFALSE, "UseMemberDecision", "use binary information from IsSignal");
DeclareOptionRef(fUseWeightedMembers=kTRUE, "UseWeightedMembers", "use weighted trees or simple average in classification from the forest");
@@ -117,6 +128,7 @@ void TMVA::MethodCommittee::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodCommittee::ProcessOptions()
{
+ // process user options
MethodBase::ProcessOptions();
}
@@ -125,7 +137,7 @@ void TMVA::MethodCommittee::InitCommittee( void )
{
// common initialisation with defaults for the Committee-Method
SetMethodName( "Committee" );
- SetMethodType( TMVA::Types::Committee );
+ SetMethodType( TMVA::Types::kCommittee );
SetTestvarName();
fNMembers = 100;
@@ -179,29 +191,29 @@ void TMVA::MethodCommittee::Train( void )
// initialize methods
switch(fMemberType) {
- case TMVA::Types::Cuts:
+ case TMVA::Types::kCuts:
method = new TMVA::MethodCuts ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::Fisher:
+ case TMVA::Types::kFisher:
method = new TMVA::MethodFisher ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::MLP:
+ case TMVA::Types::kMLP:
method = new TMVA::MethodMLP ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::TMlpANN:
+ case TMVA::Types::kTMlpANN:
method = new TMVA::MethodTMlpANN ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::CFMlpANN:
+ case TMVA::Types::kCFMlpANN:
method = new TMVA::MethodCFMlpANN ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::Likelihood:
+ case TMVA::Types::kLikelihood:
method = new TMVA::MethodLikelihood ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::HMatrix:
+ case TMVA::Types::kHMatrix:
method = new TMVA::MethodHMatrix ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::PDERS:
+ case TMVA::Types::kPDERS:
method = new TMVA::MethodPDERS ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::BDT:
+ case TMVA::Types::kBDT:
method = new TMVA::MethodBDT ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::SVM:
+ case TMVA::Types::kSVM:
method = new TMVA::MethodSVM ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::RuleFit:
+ case TMVA::Types::kRuleFit:
method = new TMVA::MethodRuleFit ( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
- case TMVA::Types::BayesClassifier:
+ case TMVA::Types::kBayesClassifier:
method = new TMVA::MethodBayesClassifier( GetJobName(), GetMethodTitle(), Data(), fMemberOption ); break;
default:
fLogger << kFATAL << "method: " << fMemberType << " does not exist" << Endl;
@@ -366,6 +378,7 @@ Double_t TMVA::MethodCommittee::Bagging( UInt_t imember )
//_______________________________________________________________________
void TMVA::MethodCommittee::WriteWeightsToStream( ostream& o ) const
{
+ // write the state of the method to an output stream
for (UInt_t imember=0; imember<GetCommittee().size(); imember++) {
o << endl;
o << "------------------------------ new member: " << imember << " ---------------" << endl;
@@ -377,6 +390,8 @@ void TMVA::MethodCommittee::WriteWeightsToStream( ostream& o ) const
//_______________________________________________________________________
void TMVA::MethodCommittee::ReadWeightsFromStream( istream& istr )
{
+ // read the state of the method from an input stream
+
// explicitly destroy objects in vector
std::vector<IMethod*>::iterator member = GetCommittee().begin();
for (; member != GetCommittee().end(); member++) delete *member;
@@ -403,29 +418,29 @@ void TMVA::MethodCommittee::ReadWeightsFromStream( istream& istr )
// initialize methods
switch(fMemberType) {
- case TMVA::Types::Cuts:
+ case TMVA::Types::kCuts:
method = new TMVA::MethodCuts ( Data(), "" ); break;
- case TMVA::Types::Fisher:
+ case TMVA::Types::kFisher:
method = new TMVA::MethodFisher ( Data(), "" ); break;
- case TMVA::Types::MLP:
+ case TMVA::Types::kMLP:
method = new TMVA::MethodMLP ( Data(), "" ); break;
- case TMVA::Types::TMlpANN:
+ case TMVA::Types::kTMlpANN:
method = new TMVA::MethodTMlpANN ( Data(), "" ); break;
- case TMVA::Types::CFMlpANN:
+ case TMVA::Types::kCFMlpANN:
method = new TMVA::MethodCFMlpANN ( Data(), "" ); break;
- case TMVA::Types::Likelihood:
+ case TMVA::Types::kLikelihood:
method = new TMVA::MethodLikelihood ( Data(), "" ); break;
- case TMVA::Types::HMatrix:
+ case TMVA::Types::kHMatrix:
method = new TMVA::MethodHMatrix ( Data(), "" ); break;
- case TMVA::Types::PDERS:
+ case TMVA::Types::kPDERS:
method = new TMVA::MethodPDERS ( Data(), "" ); break;
- case TMVA::Types::BDT:
+ case TMVA::Types::kBDT:
method = new TMVA::MethodBDT ( Data(), "" ); break;
- case TMVA::Types::SVM:
+ case TMVA::Types::kSVM:
method = new TMVA::MethodSVM ( Data(), "" ); break;
- case TMVA::Types::RuleFit:
+ case TMVA::Types::kRuleFit:
method = new TMVA::MethodRuleFit ( Data(), "" ); break;
- case TMVA::Types::BayesClassifier:
+ case TMVA::Types::kBayesClassifier:
method = new TMVA::MethodBayesClassifier( Data(), "" ); break;
default:
fLogger << kFATAL << "<ReadWeightsFromStream> fatal error: method: "
@@ -510,6 +525,7 @@ vector< Double_t > TMVA::MethodCommittee::GetVariableImportance()
//_______________________________________________________________________
Double_t TMVA::MethodCommittee::GetVariableImportance(UInt_t ivar)
{
+ // return the variable importance
vector<Double_t> relativeImportance = this->GetVariableImportance();
if (ivar < (UInt_t)relativeImportance.size()) return relativeImportance[ivar];
else fLogger << kFATAL << "<GetVariableImportance> ivar = " << ivar << " is out of range " << Endl;
diff --git a/tmva/src/MethodCuts.cxx b/tmva/src/MethodCuts.cxx
index 33d9eea77c1..47707819508 100644
--- a/tmva/src/MethodCuts.cxx
+++ b/tmva/src/MethodCuts.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodCuts.cxx,v 1.67 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodCuts.cxx,v 1.73 2006/11/17 14:59:23 stelzer Exp $
// Author: Andreas Hoecker, Matt Jachowski, Peter Speckmayer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -15,13 +15,13 @@
* Matt Jachowski <jachowski@stanford.edu> - Stanford University, USA *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
* Peter Speckmayer <speckmay@mail.cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -192,7 +192,7 @@ void TMVA::MethodCuts::InitCuts( void )
{
// default initialisation called by all constructors
SetMethodName( "Cuts" );
- SetMethodType( TMVA::Types::Cuts );
+ SetMethodType( TMVA::Types::kCuts );
SetTestvarName();
fConstrainType = kConstrainEffS;
@@ -221,7 +221,7 @@ void TMVA::MethodCuts::InitCuts( void )
fXmax = new vector<Double_t>( GetNvar() );
// get the variable specific options, first initialize default
- fFitParams = new vector<FitParameters>( GetNvar() );
+ fFitParams = new vector<EFitParameters>( GetNvar() );
for (Int_t ivar=0; ivar<GetNvar(); ivar++) (*fFitParams)[ivar] = kNotEnforced;
fRandom = new TRandom( 0 ); // set seed
@@ -280,6 +280,47 @@ TMVA::MethodCuts::~MethodCuts( void )
//_______________________________________________________________________
void TMVA::MethodCuts::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options:
+ // Method <string> Minimization method
+ // available values are: MC Monte Carlo <default>
+ // GA Genetic Algorithm
+ // SA Simulated annealing
+ //
+ // EffMethod <string> Efficiency selection method
+ // available values are: EffSel <default>
+ // EffPDF
+ //
+ // MC_NRandCuts <int> Number of random cuts to estimate the efficiency for the MC method
+ // MC_AllVarProp <string> Property of all variables for the MC method
+ // available values are: AllNotEnforced <default>
+ // AllFMax
+ // AllFMin
+ // AllFSmart
+ // AllFVerySmart
+ // MC_Var1Prop <string> Property of variable 1 for the MC method (taking precedence over the
+ // globale setting. The same values as for the global option are available. Variables 1..10 can be
+ // set this way
+ //
+ //
+ // GA_nsteps <int> Number of steps for the genetic algorithm
+ // GA_cycles <int> Number of generations for the genetic algorithm
+ // GA_popSize <int> Size of the population for the genetic algorithm
+ // GA_SC_steps <int> Number of steps for the genetic algorithm
+ // GA_SC_offsteps <int> for the genetic algorithm
+ // GA_SC_factor <float> for the genetic algorithm
+ //
+ //
+ // SA_MaxCalls <int> maximum number of calls for simulated annealing
+ // SA_TemperatureGradient <float> temperature gradient for simulated annealing
+ // SA_UseAdaptiveTemperature <bool> use of adaptive temperature for simulated annealing
+ // SA_InitialTemperature <float> initial temperature for simulated annealing
+ // SA_MinTemperature <float> minimum temperature for simulated annealing
+ // SA_Eps <int> number of epochs for simulated annealing
+ // SA_NFunLoops <int> number of loops for simulated annealing
+ // SA_NEps <int> number of epochs for simulated annealing
+
+
DeclareOptionRef(fFitMethodS="MC", "Method", "Minimization Method");
AddPreDefVal(TString("GA"));
AddPreDefVal(TString("SA"));
@@ -339,12 +380,13 @@ void TMVA::MethodCuts::DeclareOptions()
DeclareOptionRef(fSA_MinTemperature, "SA_MinTemperature", "");
DeclareOptionRef(fSA_Eps, "SA_Eps", "");
DeclareOptionRef(fSA_NFunLoops, "SA_NFunLoops", "");
- DeclareOptionRef(fSA_NEps, "SA_NEps", "");
+ DeclareOptionRef(fSA_NEps, "SA_NEps", "");
}
//_______________________________________________________________________
void TMVA::MethodCuts::ProcessOptions()
{
+ // process user options
MethodBase::ProcessOptions();
if (fFitMethodS == "MC" ) fFitMethod = kUseMonteCarlo;
@@ -377,7 +419,7 @@ void TMVA::MethodCuts::ProcessOptions()
if (fAllVars!="AllNotEnforced") { // options are specified
- FitParameters theFitP = kNotEnforced;
+ EFitParameters theFitP = kNotEnforced;
if (fAllVars == "AllNotEnforced") theFitP = kNotEnforced;
else if (fAllVars == "AllFMax" ) theFitP = kForceMax;
else if (fAllVars == "AllFMin" ) theFitP = kForceMin;
@@ -395,7 +437,7 @@ void TMVA::MethodCuts::ProcessOptions()
int maxVar = GetNvar()<=10?GetNvar():10;
for (Int_t ivar=0; ivar<maxVar; ivar++) {
- FitParameters theFitP = kNotEnforced;
+ EFitParameters theFitP = kNotEnforced;
if (fAllVarsI[ivar] == "" || fAllVarsI[ivar] == "NotEnforced") theFitP = kNotEnforced;
else if (fAllVarsI[ivar] == "FMax" ) theFitP = kForceMax;
else if (fAllVarsI[ivar] == "FMin" ) theFitP = kForceMin;
@@ -429,9 +471,9 @@ void TMVA::MethodCuts::ProcessOptions()
}
// decorrelate option will be last option, if it is specified
- if (GetPreprocessingMethod() == Types::kDecorrelated)
+ if (GetPreprocessingMethod() == Types::kDecorrelated)
fLogger << kINFO << "use decorrelated variable set" << Endl;
- else if (GetPreprocessingMethod() == Types::kPCA)
+ else if (GetPreprocessingMethod() == Types::kPCA)
fLogger << kINFO << "use principal component preprocessing" << Endl;
}
@@ -496,7 +538,7 @@ void TMVA::MethodCuts::Train( void )
const TString& varname = Data().GetInternalVarName(ivar);
- Statistics( TMVA::Types::kTrain, varname,
+ Statistics( TMVA::Types::kTraining, varname,
(*fMeanS)[ivar], (*fMeanB)[ivar],
(*fRmsS)[ivar], (*fRmsB)[ivar],
(*fXmin)[ivar], (*fXmax)[ivar] );
@@ -536,7 +578,7 @@ void TMVA::MethodCuts::Train( void )
// generate random cuts
for (Int_t ivar=0; ivar<GetNvar(); ivar++) {
- FitParameters fitParam = (*fFitParams)[ivar];
+ EFitParameters fitParam = (*fFitParams)[ivar];
if (fitParam == kForceSmart) {
if ((*fMeanS)[ivar] > (*fMeanB)[ivar]) fitParam = kForceMax;
@@ -643,13 +685,13 @@ void TMVA::MethodCuts::Train( void )
timer1.DrawProgressBar( cycle );
// ---- perform series of fits to achieve best convergence
-
+
// "m_ga_spread" times the number of variables
TMVA::GeneticCuts ga( fGA_popSize, ranges, this );
-
+
ga.CalculateFitness();
ga.GetGeneticPopulation().TrimPopulation();
-
+
do {
ga.Init();
ga.CalculateFitness();
@@ -728,9 +770,9 @@ void TMVA::MethodCuts::Train( void )
if (fBinaryTreeB != 0) { delete fBinaryTreeB; fBinaryTreeB = 0; }
}
-void TMVA::MethodCuts::Test( TTree* theTestTree )
+void TMVA::MethodCuts::Test( TTree* )
{
- if (theTestTree == 0); // dummy call
+ // not used
}
//_______________________________________________________________________
@@ -998,6 +1040,7 @@ void TMVA::MethodCuts::WriteWeightsToStream( ostream & o ) const
//_______________________________________________________________________
void TMVA::MethodCuts::ReadWeightsFromStream( istream& istr )
{
+ // read the cuts from stream
TString dummy;
UInt_t dummyInt;
diff --git a/tmva/src/MethodFisher.cxx b/tmva/src/MethodFisher.cxx
index 127e14843df..3f061910b52 100644
--- a/tmva/src/MethodFisher.cxx
+++ b/tmva/src/MethodFisher.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodFisher.cxx,v 1.53 2006/11/14 15:22:54 helgevoss Exp $
+// @(#)root/tmva $Id: MethodFisher.cxx,v 1.58 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Xavier Prudent, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -17,13 +17,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -171,7 +171,7 @@ void TMVA::MethodFisher::InitFisher( void )
{
// default initialisation called by all constructors
SetMethodName( "Fisher" );
- SetMethodType( TMVA::Types::Fisher );
+ SetMethodType( TMVA::Types::kFisher );
SetTestvarName();
fMeanMatx = 0;
@@ -201,6 +201,7 @@ void TMVA::MethodFisher::DeclareOptions()
void TMVA::MethodFisher::ProcessOptions()
{
+ // process user options
MethodBase::ProcessOptions();
if (fTheMethod == "Fisher" ) fFisherMethod = kFisher;
@@ -410,11 +411,19 @@ void TMVA::MethodFisher::GetFisherCoeff( void )
}
TMatrixD invCov( *theMat );
- if ( TMath::Abs(invCov.Determinant()) < 10E-14 ) {
- fLogger << kFATAL << "<GetFisherCoeff> matrix is singular,"
- << " did you use the variables that are linear combinations ???"
- << GetFisherMethod() << Endl;
- }
+ if ( TMath::Abs(invCov.Determinant()) < 10E-24 ) {
+ fLogger << kWARNING << "<GetFisherCoeff> matrix is almost singular with deterninant="
+ << TMath::Abs(invCov.Determinant())
+ << " did you use the variables that are linear combinations or highly correlated ???"
+ << Endl;
+ }
+ if ( TMath::Abs(invCov.Determinant()) < 10E-120 ) {
+ fLogger << kFATAL << "<GetFisherCoeff> matrix is singular with determinant="
+ << TMath::Abs(invCov.Determinant())
+ << " did you use the variables that are linear combinations ???"
+ << Endl;
+ }
+
invCov.Invert();
// apply rescaling factor
@@ -440,7 +449,7 @@ void TMVA::MethodFisher::GetFisherCoeff( void )
fF0 = 0.0;
for(ivar=0; ivar<GetNvar(); ivar++){
fF0 += (*fFisherCoeff)[ivar]*((*fMeanMatx)(ivar, 0) + (*fMeanMatx)(ivar, 1));
- }
+ }
fF0 /= -2.0;
}
diff --git a/tmva/src/MethodHMatrix.cxx b/tmva/src/MethodHMatrix.cxx
index 36ff36a978a..ace282ddcdb 100644
--- a/tmva/src/MethodHMatrix.cxx
+++ b/tmva/src/MethodHMatrix.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodHMatrix.cxx,v 1.31 2006/11/14 16:00:56 helgevoss Exp $
+// @(#)root/tmva $Id: MethodHMatrix.cxx,v 1.36 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Xavier Prudent, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -90,7 +90,7 @@ void TMVA::MethodHMatrix::InitHMatrix( void )
{
// default initialisation called by all constructors
SetMethodName( "HMatrix" );
- SetMethodType( TMVA::Types::HMatrix );
+ SetMethodType( TMVA::Types::kHMatrix );
SetTestvarName();
fNormaliseInputVars = kTRUE;
@@ -122,6 +122,7 @@ void TMVA::MethodHMatrix::DeclareOptions()
void TMVA::MethodHMatrix::ProcessOptions()
{
+ // process user options
MethodBase::ProcessOptions();
}
@@ -137,7 +138,7 @@ void TMVA::MethodHMatrix::Train( void )
Double_t meanS, meanB, rmsS, rmsB, xmin, xmax;
for (Int_t ivar=0; ivar<GetNvar(); ivar++) {
- Statistics( TMVA::Types::kTrain, (*fInputVars)[ivar],
+ Statistics( TMVA::Types::kTraining, (*fInputVars)[ivar],
meanS, meanB, rmsS, rmsB, xmin, xmax, fNormaliseInputVars );
(*fVecMeanS)(ivar) = meanS;
@@ -148,17 +149,33 @@ void TMVA::MethodHMatrix::Train( void )
this->ComputeCovariance( kTRUE, fInvHMatrixS );
this->ComputeCovariance( kFALSE, fInvHMatrixB );
- if ( TMath::Abs(fInvHMatrixS->Determinant()) < 10E-14 ) {
- fLogger << kFATAL << "<Train> matrix is singular,"
- << " did you use the variables that are linear combinations ???"
+ if ( TMath::Abs(fInvHMatrixS->Determinant()) < 10E-24 ) {
+ fLogger << kWARNING << "<Train> H-matrix S is almost singular with deterinant= "
+ << TMath::Abs(fInvHMatrixS->Determinant())
+ << " did you use the variables that are linear combinations or highly correlated ???"
<< Endl;
}
- if ( TMath::Abs(fInvHMatrixB->Determinant()) < 10E-14 ) {
- fLogger << kFATAL << "<Train> matrix is singular,"
- << " did you use the variables that are linear combinations ???"
+ if ( TMath::Abs(fInvHMatrixB->Determinant()) < 10E-24 ) {
+ fLogger << kWARNING << "<Train> H-matrix B is almost singular with deterinant= "
+ << TMath::Abs(fInvHMatrixB->Determinant())
+ << " did you use the variables that are linear combinations or highly correlated ???"
<< Endl;
}
+ if ( TMath::Abs(fInvHMatrixS->Determinant()) < 10E-120 ) {
+ fLogger << kFATAL << "<Train> H-matrix S is singular with deterinant= "
+ << TMath::Abs(fInvHMatrixS->Determinant())
+ << " did you use the variables that are linear combinations ???"
+ << Endl;
+ }
+ if ( TMath::Abs(fInvHMatrixB->Determinant()) < 10E-120 ) {
+ fLogger << kFATAL << "<Train> H-matrix B is singular with deterinant= "
+ << TMath::Abs(fInvHMatrixB->Determinant())
+ << " did you use the variables that are linear combinations ???"
+ << Endl;
+ }
+
+
// invert matrix
@@ -229,7 +246,7 @@ Double_t TMVA::MethodHMatrix::GetMvaValue()
}
//_______________________________________________________________________
-Double_t TMVA::MethodHMatrix::GetChi2( TMVA::Event *e, Types::SBType type ) const
+Double_t TMVA::MethodHMatrix::GetChi2( TMVA::Event *e, Types::ESBType type ) const
{
// compute chi2-estimator for event according to type (signal/background)
@@ -261,7 +278,7 @@ Double_t TMVA::MethodHMatrix::GetChi2( TMVA::Event *e, Types::SBType type ) con
}
//_______________________________________________________________________
-Double_t TMVA::MethodHMatrix::GetChi2( Types::SBType type ) const
+Double_t TMVA::MethodHMatrix::GetChi2( Types::ESBType type ) const
{
// compute chi2-estimator for event according to type (signal/background)
diff --git a/tmva/src/MethodLikelihood.cxx b/tmva/src/MethodLikelihood.cxx
index e12d3845e94..4e1787ca94f 100644
--- a/tmva/src/MethodLikelihood.cxx
+++ b/tmva/src/MethodLikelihood.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodLikelihood.cxx,v 1.51 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodLikelihood.cxx,v 1.55 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -186,7 +186,7 @@ void TMVA::MethodLikelihood::InitLik( void )
fPDFBgd = NULL;
SetMethodName( "Likelihood" );
- SetMethodType( TMVA::Types::Likelihood );
+ SetMethodType( TMVA::Types::kLikelihood );
SetTestvarName();
fEpsilon = 1e-5;
@@ -207,6 +207,17 @@ void TMVA::MethodLikelihood::InitLik( void )
//_______________________________________________________________________
void TMVA::MethodLikelihood::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options:
+ // Spline <int> spline used to interpolate reference histograms
+ // available values are: 0, 1, 2 <default>, 3, 5
+ //
+ // NSmooth <int> how often the input histos are smoothed
+ // NAvEvtPerBin <int> minimum average number of events per PDF bin (less trigger warning)
+ // TransformOutput <bool> transform (often strongly peaked) likelihood output through sigmoid inversion
+
+
+
DeclareOptionRef(fSpline=2,"Spline","spline used to interpolate reference histograms");
AddPreDefVal(0); // take histogram
AddPreDefVal(1); // linear interpolation between bins
@@ -226,6 +237,7 @@ void TMVA::MethodLikelihood::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodLikelihood::ProcessOptions()
{
+ // process user options
MethodBase::ProcessOptions();
if (fSpline == 0) fSmoothMethod = TMVA::PDF::kSpline0;
@@ -239,9 +251,9 @@ void TMVA::MethodLikelihood::ProcessOptions()
}
// decorrelate option will be last option, if it is specified
- if (GetPreprocessingMethod() == Types::kDecorrelated)
+ if (GetPreprocessingMethod() == Types::kDecorrelated)
fLogger << kINFO << "use decorrelated variable set" << Endl;
- else if (GetPreprocessingMethod() == Types::kPCA)
+ else if (GetPreprocessingMethod() == Types::kPCA)
fLogger << kINFO << "use principal component preprocessing" << Endl;
}
@@ -563,8 +575,11 @@ void TMVA::MethodLikelihood::ReadWeightsFromStream( istream& istr )
// find corresponding variable index and cache it (to spead up likelihood evaluation)
for (Int_t ivar=0; ivar<GetNvar(); ivar++) {
if (hname.Contains( (*fInputVars)[ivar] )) {
- if (hname.Contains("_sig_")) (*fIndexSig)[ivar] = fSigPDFHist->GetEntries()-1;
- else if (hname.Contains("_bgd_")) (*fIndexBgd)[ivar] = fBgdPDFHist->GetEntries()-1;
+// if (hname.Contains("_sig_")) (*fIndexSig)[ivar] = fSigPDFHist->GetEntries()-1;
+// else if (hname.Contains("_bgd_")) (*fIndexBgd)[ivar] = fBgdPDFHist->GetEntries()-1;
+// to be backward compatible to ROOT 4.02
+ if (hname.Contains("_sig_")) (*fIndexSig)[ivar] = fSigPDFHist->GetSize()-1;
+ else if (hname.Contains("_bgd_")) (*fIndexBgd)[ivar] = fBgdPDFHist->GetSize()-1;
}
}
}
diff --git a/tmva/src/MethodMLP.cxx b/tmva/src/MethodMLP.cxx
index 82eaeb72776..292d633982d 100644
--- a/tmva/src/MethodMLP.cxx
+++ b/tmva/src/MethodMLP.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodMLP.cxx,v 1.29 2006/11/14 23:02:57 stelzer Exp $
+// @(#)root/tmva $Id: MethodMLP.cxx,v 1.32 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Matt Jachowski
/**********************************************************************************
@@ -113,13 +113,29 @@ void TMVA::MethodMLP::InitMLP()
{
// default initializations
SetMethodName( "MLP" );
- SetMethodType( TMVA::Types::MLP );
+ SetMethodType( TMVA::Types::kMLP );
SetTestvarName();
}
//_______________________________________________________________________
void TMVA::MethodMLP::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options:
+ // TrainingMethod <string> Training method
+ // available values are: BP Back-Propagation <default>
+ // GA Genetic Algorithm (takes a LONG time)
+ //
+ // LearningRate <float> NN learning rate parameter
+ // DecayRate <float> Decay rate for learning parameter
+ // TestRate <int> Test for overtraining performed at each #th epochs
+ //
+ // BPMode <string> Back-propagation learning mode
+ // available values are: sequential <default>
+ // batch
+ //
+ // BatchSize <int> Batch size: number of events/batch, only set if in Batch Mode, -1 for BatchSize=number_of_events
+
DeclareOptionRef(fTrainMethodS="BP", "TrainingMethod",
"Train with Back-Propagation (BP) or Genetic Algorithm (GA) (takes a LONG time)");
AddPreDefVal(TString("BP"));
@@ -141,6 +157,7 @@ void TMVA::MethodMLP::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodMLP::ProcessOptions()
{
+ // process user options
MethodANNBase::ProcessOptions();
if (fTrainMethodS == "BP") fTrainingMethod = kBP;
@@ -153,7 +170,6 @@ void TMVA::MethodMLP::ProcessOptions()
//______________________________________________________________________________
void TMVA::MethodMLP::InitializeLearningRates()
{
-
// initialize learning rates of synapses, used only by backpropagation
TSynapse *synapse;
Int_t numSynapses = fSynapses->GetEntriesFast();
@@ -164,12 +180,12 @@ void TMVA::MethodMLP::InitializeLearningRates()
}
//______________________________________________________________________________
-Double_t TMVA::MethodMLP::CalculateEstimator( TMVA::Types::TreeType treeType )
+Double_t TMVA::MethodMLP::CalculateEstimator( TMVA::Types::ETreeType treeType )
{
// calculate the estimator that training is attempting to minimize
- Int_t nEvents = ( (treeType == TMVA::Types::kTest ) ? Data().GetNEvtTest() :
- (treeType == TMVA::Types::kTrain) ? Data().GetNEvtTrain() : -1 );
+ Int_t nEvents = ( (treeType == TMVA::Types::kTesting ) ? Data().GetNEvtTest() :
+ (treeType == TMVA::Types::kTraining) ? Data().GetNEvtTrain() : -1 );
// sanity check
if (nEvents <=0)
@@ -180,8 +196,10 @@ Double_t TMVA::MethodMLP::CalculateEstimator( TMVA::Types::TreeType treeType )
// loop over all training events
for (Int_t i = 0; i < nEvents; i++) {
- if (treeType == TMVA::Types::kTest ) ReadTestEvent(i);
- else ReadTrainingEvent(i);
+ if (treeType == TMVA::Types::kTesting )
+ ReadTestEvent(i);
+ else
+ ReadTrainingEvent(i);
Double_t desired = GetDesiredOutput();
ForceNetworkInputs();
@@ -236,8 +254,8 @@ void TMVA::MethodMLP::BackPropagationMinimize(Int_t nEpochs)
// monitor convergence of training and control sample
if ((i+1)%fTestRate == 0) {
- Double_t trainE = CalculateEstimator( TMVA::Types::kTrain ); // estimator for training sample
- Double_t testE = CalculateEstimator( TMVA::Types::kTest ); // estimator for test samplea
+ Double_t trainE = CalculateEstimator( TMVA::Types::kTraining ); // estimator for training sample
+ Double_t testE = CalculateEstimator( TMVA::Types::kTesting ); // estimator for test samplea
fEstimatorHistTrain->Fill( i+1, trainE );
fEstimatorHistTest ->Fill( i+1, testE );
}
@@ -579,6 +597,7 @@ void TMVA::MethodMLP::AdjustSynapseWeights()
//______________________________________________________________________________
void TMVA::MethodMLP::MinuitMinimize()
{
+ // minimize using Minuit
fNumberOfWeights = fSynapses->GetEntriesFast();
TFitter* tfitter = new TFitter( fNumberOfWeights );
diff --git a/tmva/src/MethodPDERS.cxx b/tmva/src/MethodPDERS.cxx
index 398e8cb54e0..863a628d2c4 100644
--- a/tmva/src/MethodPDERS.cxx
+++ b/tmva/src/MethodPDERS.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodPDERS.cxx,v 1.39 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodPDERS.cxx,v 1.43 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Yair Mahalalel, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Yair Mahalalel <Yair.Mahalalel@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -123,7 +123,7 @@ void TMVA::MethodPDERS::InitPDERS( void )
{
// default initialisation routine called by all constructors
SetMethodName( "PDERS" );
- SetMethodType( TMVA::Types::PDERS );
+ SetMethodType( TMVA::Types::kPDERS );
SetTestvarName();
fBinaryTreeS = fBinaryTreeB = NULL;
@@ -160,6 +160,36 @@ TMVA::MethodPDERS::~MethodPDERS( void )
//_______________________________________________________________________
void TMVA::MethodPDERS::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options:
+ // VolumeRangeMode <string> Method to determine volume range
+ // available values are: MinMax <default>
+ // Unscaled
+ // RMS
+ // Adaptive
+ //
+ // KernelEstimator <string> Kernel estimation function
+ // available values are: Box <default>
+ // Sphere
+ // Teepee
+ // Gauss
+ // Sinc3
+ // Sinc5
+ // Sinc7
+ // Sinc9
+ // Sinc11
+ // Lanczos2
+ // Lanczos3
+ // Lanczos5
+ // Lanczos8
+ //
+ // DeltaFrac <float> Ratio of #EventsMin/#EventsMax for MinMax and RMS volume range
+ // NEventsMin <int> Minimum number of events for adaptive volume range
+ // NEventsMax <int> Maximum number of events for adaptive volume range
+ // MaxVIterations <int> Maximum number of iterations for adaptive volume range
+ // InitialScale <float> Initial scale for adaptive volume range
+ // GaussSigma <float> Width with respect to the volume size of Gaussian kernel estimator
+
DeclareOptionRef(fVolumeRange="MinMax", "VolumeRangeMode", "Method to determine volume range");
AddPreDefVal(TString("Unscaled"));
AddPreDefVal(TString("MinMax"));
@@ -181,7 +211,7 @@ void TMVA::MethodPDERS::DeclareOptions()
AddPreDefVal(TString("Lanczos5"));
AddPreDefVal(TString("Lanczos8"));
- DeclareOptionRef(fDeltaFrac , "DeltaFrac", "nEventsMin/Max for minmax and rms vaolume range");
+ DeclareOptionRef(fDeltaFrac , "DeltaFrac", "nEventsMin/Max for minmax and rms volume range");
DeclareOptionRef(fNEventsMin , "NEventsMin", "nEventsMin for adaptive volume range");
DeclareOptionRef(fNEventsMax , "NEventsMax", "nEventsMax for adaptive volume range");
DeclareOptionRef(fMaxVIterations, "MaxVIterations", "MaxVIterations for adaptive volume range");
@@ -192,36 +222,38 @@ void TMVA::MethodPDERS::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodPDERS::ProcessOptions()
{
+ // process the options specified by the user
+
MethodBase::ProcessOptions();
fVRangeMode = TMVA::MethodPDERS::kUnsupported;
- if (fVolumeRange == "MinMax" ) fVRangeMode = TMVA::MethodPDERS::kMinMax;
- else if (fVolumeRange == "RMS" ) fVRangeMode = TMVA::MethodPDERS::kRMS;
- else if (fVolumeRange == "Adaptive" ) fVRangeMode = TMVA::MethodPDERS::kAdaptive;
- else if (fVolumeRange == "Unscaled" ) fVRangeMode = TMVA::MethodPDERS::kUnscaled;
+ if (fVolumeRange == "MinMax" ) fVRangeMode = TMVA::MethodPDERS::kMinMax;
+ else if (fVolumeRange == "RMS" ) fVRangeMode = TMVA::MethodPDERS::kRMS;
+ else if (fVolumeRange == "Adaptive" ) fVRangeMode = TMVA::MethodPDERS::kAdaptive;
+ else if (fVolumeRange == "Unscaled" ) fVRangeMode = TMVA::MethodPDERS::kUnscaled;
else {
fLogger << kFATAL << "VolumeRangeMode parameter '" << fVolumeRange << "' unknown" << Endl;
}
- if (fKernelString == "Box" ) fKernelEstimator = TMVA::MethodPDERS::kBox;
- else if (fKernelString == "Sphere" ) fKernelEstimator = TMVA::MethodPDERS::kSphere;
- else if (fKernelString == "Teepee" ) fKernelEstimator = TMVA::MethodPDERS::kTeepee;
- else if (fKernelString == "Gauss" ) fKernelEstimator = TMVA::MethodPDERS::kGauss;
- else if (fKernelString == "Sinc3" ) fKernelEstimator = TMVA::MethodPDERS::kSinc3;
- else if (fKernelString == "Sinc5" ) fKernelEstimator = TMVA::MethodPDERS::kSinc5;
- else if (fKernelString == "Sinc7" ) fKernelEstimator = TMVA::MethodPDERS::kSinc7;
- else if (fKernelString == "Sinc9" ) fKernelEstimator = TMVA::MethodPDERS::kSinc9;
- else if (fKernelString == "Sinc11" ) fKernelEstimator = TMVA::MethodPDERS::kSinc11;
- else if (fKernelString == "Lanczos2" ) fKernelEstimator = TMVA::MethodPDERS::kLanczos2;
- else if (fKernelString == "Lanczos3" ) fKernelEstimator = TMVA::MethodPDERS::kLanczos3;
- else if (fKernelString == "Lanczos5" ) fKernelEstimator = TMVA::MethodPDERS::kLanczos5;
- else if (fKernelString == "Lanczos8" ) fKernelEstimator = TMVA::MethodPDERS::kLanczos8;
+ if (fKernelString == "Box" ) fKernelEstimator = TMVA::MethodPDERS::kBox;
+ else if (fKernelString == "Sphere" ) fKernelEstimator = TMVA::MethodPDERS::kSphere;
+ else if (fKernelString == "Teepee" ) fKernelEstimator = TMVA::MethodPDERS::kTeepee;
+ else if (fKernelString == "Gauss" ) fKernelEstimator = TMVA::MethodPDERS::kGauss;
+ else if (fKernelString == "Sinc3" ) fKernelEstimator = TMVA::MethodPDERS::kSinc3;
+ else if (fKernelString == "Sinc5" ) fKernelEstimator = TMVA::MethodPDERS::kSinc5;
+ else if (fKernelString == "Sinc7" ) fKernelEstimator = TMVA::MethodPDERS::kSinc7;
+ else if (fKernelString == "Sinc9" ) fKernelEstimator = TMVA::MethodPDERS::kSinc9;
+ else if (fKernelString == "Sinc11" ) fKernelEstimator = TMVA::MethodPDERS::kSinc11;
+ else if (fKernelString == "Lanczos2" ) fKernelEstimator = TMVA::MethodPDERS::kLanczos2;
+ else if (fKernelString == "Lanczos3" ) fKernelEstimator = TMVA::MethodPDERS::kLanczos3;
+ else if (fKernelString == "Lanczos5" ) fKernelEstimator = TMVA::MethodPDERS::kLanczos5;
+ else if (fKernelString == "Lanczos8" ) fKernelEstimator = TMVA::MethodPDERS::kLanczos8;
else {
fLogger << kFATAL << "KernelEstimator parameter '" << fKernelString << "' unknown" << Endl;
}
- // TODO: Add parameter validation
+ // TODO: Add parameter validation
fLogger << kVERBOSE << "interpreted option string: vRangeMethod: '"
<< (const char*)((fVRangeMode == kMinMax) ? "MinMax" :
@@ -438,7 +470,7 @@ Float_t TMVA::MethodPDERS::RScalc( const TMVA::Event& e )
delete lb;
delete ub;
}
- else if (fVRangeMode == kAdaptive) { // adaptive volume
+ else if (fVRangeMode == kAdaptive) { // adaptive volume
// -----------------------------------------------------------------------
@@ -554,7 +586,7 @@ Float_t TMVA::MethodPDERS::RScalc( const TMVA::Event& e )
if (countS < 1e-20) return 0.0;
Float_t r = countB*fScaleB/(countS*fScaleS);
- return 1.0/(r + 1.0); // TODO: propagate errors from here
+ return 1.0/(r + 1.0); // TODO: propagate errors from here
}
//_______________________________________________________________________
@@ -577,7 +609,7 @@ Double_t TMVA::MethodPDERS::KernelEstimate( const TMVA::Event & event,
// always working within the hyperelipsoid, except for when we don't
// note that rejection ratio goes to 1 as nvar goes to infinity
- if (normalized_distance > 1 && fKernelEstimator != kBox) continue;
+ if (normalized_distance > 1 && fKernelEstimator != kBox) continue;
pdfSum += ApplyKernelFunction (normalized_distance) * (*iev)->GetWeight();
}
@@ -587,6 +619,8 @@ Double_t TMVA::MethodPDERS::KernelEstimate( const TMVA::Event & event,
//_______________________________________________________________________
Double_t TMVA::MethodPDERS::ApplyKernelFunction (Double_t normalized_distance)
{
+ // from the normalized euclidean distance calculate the distance
+ // for a certain kernel
switch (fKernelEstimator) {
case kBox:
case kSphere:
@@ -605,8 +639,8 @@ Double_t TMVA::MethodPDERS::ApplyKernelFunction (Double_t normalized_distance)
case kSinc11: {
Double_t side_crossings = 2 + ((int) fKernelEstimator) - ((int) kSinc3);
return NormSinc (side_crossings * normalized_distance);
- }
- break;
+ }
+ break;
case kLanczos2:
return LanczosFilter (2, normalized_distance);
break;
@@ -626,25 +660,25 @@ Double_t TMVA::MethodPDERS::ApplyKernelFunction (Double_t normalized_distance)
return 0;
}
-
+
//_______________________________________________________________________
Double_t TMVA::MethodPDERS::KernelNormalization (Double_t pdf)
{
- // Calculating the normalization factor only once (might need a reset at some point. Can the method be restarted with different params?)
+ // Calculating the normalization factor only once (might need a reset at some point. Can the method be restarted with different params?)
- static Double_t ret = 1.; // Caching jammed to disable function. It's not really useful afterall, badly implemented and untested :-)
+ static Double_t ret = 1.; // Caching jammed to disable function. It's not really useful afterall, badly implemented and untested :-)
- if (ret != 0.)
- return ret*pdf;
+ if (ret != 0.)
+ return ret*pdf;
- // We first normalize by the volume of the hypersphere.
- switch (fKernelEstimator) {
+ // We first normalize by the volume of the hypersphere.
+ switch (fKernelEstimator) {
case kBox:
case kSphere:
ret = 1.;
break;
case kTeepee:
- ret = (GetNvar() * (GetNvar() + 1) * TMath::Gamma (((Double_t) GetNvar()) / 2.)) /
+ ret = (GetNvar() * (GetNvar() + 1) * TMath::Gamma (((Double_t) GetNvar()) / 2.)) /
( TMath::Power (2., (Double_t) GetNvar() + 1) * TMath::Power (TMath::Pi(), ((Double_t) GetNvar()) / 2.));
break;
case kGauss:
@@ -665,63 +699,65 @@ Double_t TMVA::MethodPDERS::KernelNormalization (Double_t pdf)
break;
default:
fLogger << kFATAL << "Kernel estimation function unsupported. Enumerator is " << fKernelEstimator << Endl;
- }
- // Normalizing by the full volume
- ret *= ( TMath::Power (2., GetNvar()) * TMath::Gamma (1 + (((Double_t) GetNvar()) / 2.)) ) /
+ }
+ // Normalizing by the full volume
+ ret *= ( TMath::Power (2., GetNvar()) * TMath::Gamma (1 + (((Double_t) GetNvar()) / 2.)) ) /
TMath::Power (TMath::Pi(), ((Double_t) GetNvar()) / 2.);
- return ret*pdf;
+ return ret*pdf;
}
//_______________________________________________________________________
-Double_t TMVA::MethodPDERS::GetNormalizedDistance ( const TMVA::Event &base_event,
+Double_t TMVA::MethodPDERS::GetNormalizedDistance ( const TMVA::Event &base_event,
const TMVA::Event &sample_event,
Double_t *dim_normalization) {
- // We use Euclidian metric here. Might not be best or most efficient.
- Double_t ret=0;
- for (Int_t ivar=0; ivar<GetNvar(); ivar++) {
- Double_t dist = dim_normalization[ivar] * (sample_event.GetVal(ivar) - base_event.GetVal(ivar));
- ret += dist*dist;
- }
- return TMath::Sqrt (ret);
+ // We use Euclidian metric here. Might not be best or most efficient.
+ Double_t ret=0;
+ for (Int_t ivar=0; ivar<GetNvar(); ivar++) {
+ Double_t dist = dim_normalization[ivar] * (sample_event.GetVal(ivar) - base_event.GetVal(ivar));
+ ret += dist*dist;
+ }
+ return TMath::Sqrt (ret);
}
//_______________________________________________________________________
-Double_t TMVA::MethodPDERS::NormSinc (Double_t x) {
-
- if (x < 10e-10 && x > -10e-10) {
- return 1; // Poor man's l'Hopital
- }
-
- Double_t pix = TMath::Pi() * x;
- Double_t sinc = TMath::Sin(pix) / pix;
- Double_t ret;
-
- if (GetNvar() % 2)
- ret = TMath::Power (sinc, GetNvar());
- else
- ret = TMath::Abs (sinc) * TMath::Power (sinc, GetNvar() - 1);
-
- return ret;
+Double_t TMVA::MethodPDERS::NormSinc (Double_t x)
+{
+ // NormSinc
+ if (x < 10e-10 && x > -10e-10) {
+ return 1; // Poor man's l'Hopital
+ }
+
+ Double_t pix = TMath::Pi() * x;
+ Double_t sinc = TMath::Sin(pix) / pix;
+ Double_t ret;
+
+ if (GetNvar() % 2)
+ ret = TMath::Power (sinc, GetNvar());
+ else
+ ret = TMath::Abs (sinc) * TMath::Power (sinc, GetNvar() - 1);
+
+ return ret;
}
//_______________________________________________________________________
-Double_t TMVA::MethodPDERS::LanczosFilter (Int_t level, Double_t x) {
-
- if (x < 10e-10 && x > -10e-10) {
- return 1; // Poor man's l'Hopital
- }
-
- Double_t pix = TMath::Pi() * x;
- Double_t pixtimesn = pix * ((Double_t) level);
- Double_t lanczos = (TMath::Sin(pix) / pix) * (TMath::Sin(pixtimesn) / pixtimesn);
- Double_t ret;
-
- if (GetNvar() % 2)
- ret = TMath::Power (lanczos, GetNvar());
- else
- ret = TMath::Abs (lanczos) * TMath::Power (lanczos, GetNvar() - 1);
-
- return ret;
+Double_t TMVA::MethodPDERS::LanczosFilter (Int_t level, Double_t x)
+{
+ // Lanczos Filter
+ if (x < 10e-10 && x > -10e-10) {
+ return 1; // Poor man's l'Hopital
+ }
+
+ Double_t pix = TMath::Pi() * x;
+ Double_t pixtimesn = pix * ((Double_t) level);
+ Double_t lanczos = (TMath::Sin(pix) / pix) * (TMath::Sin(pixtimesn) / pixtimesn);
+ Double_t ret;
+
+ if (GetNvar() % 2)
+ ret = TMath::Power (lanczos, GetNvar());
+ else
+ ret = TMath::Abs (lanczos) * TMath::Power (lanczos, GetNvar() - 1);
+
+ return ret;
}
//_______________________________________________________________________
diff --git a/tmva/src/MethodRuleFit.cxx b/tmva/src/MethodRuleFit.cxx
index 8d707269a05..fd634730eda 100644
--- a/tmva/src/MethodRuleFit.cxx
+++ b/tmva/src/MethodRuleFit.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodRuleFit.cxx,v 1.40 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodRuleFit.cxx,v 1.43 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -16,7 +16,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -87,6 +87,30 @@ TMVA::MethodRuleFit::~MethodRuleFit( void )
//_______________________________________________________________________
void TMVA::MethodRuleFit::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options:
+ // GDTau <float> gradient-directed path: fit threshhold
+ // GDStep <float> gradient-directed path: step size
+ // GDNSteps <float> gradient-directed path: number of steps
+ // GDErrNsigma <float> threshold for error-rate
+ // MinImp <float> minimum rule importance accepted
+ // nEventsMin <float> minimum number of events in a leaf node
+ // nTrees <float> number of trees in forest.
+ // SampleFraction <float> fraction of events used to train each tree
+ // nCuts <float> number of steps during node cut optimisation
+ // RuleMaxDist <float> max distance allowed between equal rules
+ //
+ // SeparationType <string> separation criterion for node splitting
+ // available values are: GiniIndex <default>
+ // MisClassificationError
+ // CrossEntropy
+ // SDivSqrtSPlusB
+ //
+ // Model <string> model to be used
+ // available values are: ModRuleLinear <default>
+ // ModRule
+ // ModLinear
+
DeclareOptionRef(fGDTau=0.0, "GDTau", "gradient-directed path: fit threshhold");
DeclareOptionRef(fGDPathStep=0.01, "GDStep", "gradient-directed path: step size");
DeclareOptionRef(fGDNPathSteps=100, "GDNSteps", "gradient-directed path: number of steps");
@@ -113,6 +137,7 @@ void TMVA::MethodRuleFit::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodRuleFit::ProcessOptions()
{
+ // process the options specified by the user
MethodBase::ProcessOptions();
if (fSepTypeS == "misclassificationerror") fSepType = new TMVA::MisClassificationError();
@@ -135,6 +160,7 @@ void TMVA::MethodRuleFit::ProcessOptions()
//_______________________________________________________________________
void TMVA::MethodRuleFit::InitMonitorNtuple()
{
+ // initialize the monitoring ntuple
fMonitorNtuple= new TTree("MonitorNtuple_RuleFit","RuleFit variables");
fMonitorNtuple->Branch("importance",&fNTImportance,"importance/D");
fMonitorNtuple->Branch("support",&fNTSupport,"support/D");
@@ -154,7 +180,7 @@ void TMVA::MethodRuleFit::InitRuleFit( void )
{
// default initialisation
SetMethodName( "RuleFit" );
- SetMethodType( TMVA::Types::RuleFit );
+ SetMethodType( TMVA::Types::kRuleFit );
SetTestvarName();
}
@@ -191,6 +217,7 @@ void TMVA::MethodRuleFit::InitEventSample( void )
//_______________________________________________________________________
void TMVA::MethodRuleFit::BuildTree( TMVA::DecisionTree *dt, std::vector< TMVA::Event *> & el )
{
+ // build the decision tree
if (dt==0) return;
dt->BuildTree(el);
}
@@ -298,13 +325,14 @@ const TMVA::Ranking* TMVA::MethodRuleFit::CreateRanking()
//_______________________________________________________________________
void TMVA::MethodRuleFit::WriteWeightsToStream( ostream & o ) const
{
+ // write the rules to an ostream
fRuleFit.GetRuleEnsemble().PrintRaw(o);
}
//_______________________________________________________________________
void TMVA::MethodRuleFit::ReadWeightsFromStream( istream & istr )
{
- // read rules from stream
+ // read rules from an istream
fRuleFit.GetRuleEnsemblePtr()->ReadRaw(istr);
}
diff --git a/tmva/src/MethodSVM.cxx b/tmva/src/MethodSVM.cxx
index 2c8055ea8d7..7d713ef68a0 100644
--- a/tmva/src/MethodSVM.cxx
+++ b/tmva/src/MethodSVM.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodSVM.cxx,v 1.16 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodSVM.cxx,v 1.19 2006/11/17 00:21:35 stelzer Exp $
// Author: Marcin ....
/**********************************************************************************
@@ -17,7 +17,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -60,15 +60,18 @@ void TMVA::MethodSVM::InitSVM( void )
{
// default initialisation
SetMethodName( "SVM" );
- SetMethodType( TMVA::Types::SVM );
+ SetMethodType( TMVA::Types::kSVM );
SetTestvarName();
}
void TMVA::MethodSVM::DeclareOptions()
-{}
+{
+ // declare options available for this method
+}
void TMVA::MethodSVM::ProcessOptions()
{
+ // evaluate options
MethodBase::ProcessOptions();
}
@@ -90,12 +93,14 @@ void TMVA::MethodSVM::Train( void )
//_______________________________________________________________________
void TMVA::MethodSVM::WriteWeightsToStream( ostream & o ) const
{
+ // Write configuration to output stream
o << "whatever" << endl;
}
//_______________________________________________________________________
void TMVA::MethodSVM::ReadWeightsFromStream( istream & istr )
{
+ // Read configuration from input stream
if (istr.eof());
}
diff --git a/tmva/src/MethodTMlpANN.cxx b/tmva/src/MethodTMlpANN.cxx
index 4b556100180..298c2fb8ab8 100644
--- a/tmva/src/MethodTMlpANN.cxx
+++ b/tmva/src/MethodTMlpANN.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodTMlpANN.cxx,v 1.31 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodTMlpANN.cxx,v 1.34 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
* Project: TMVA - a Root-integrated toolkit for multivariate data analysis *
@@ -12,13 +12,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -100,7 +100,7 @@ void TMVA::MethodTMlpANN::InitTMlpANN( void )
{
// default initialisations
SetMethodName( "TMlpANN" );
- SetMethodType( TMVA::Types::TMlpANN );
+ SetMethodType( TMVA::Types::kTMlpANN );
SetTestvarName();
}
@@ -157,6 +157,16 @@ void TMVA::MethodTMlpANN::CreateMLPOptions( TString layerSpec )
//_______________________________________________________________________
void TMVA::MethodTMlpANN::DeclareOptions()
{
+ // define the options (their key words) that can be set in the option string
+ // know options:
+ // NCycles <integer> Number of training cycles (too many cycles could overtrain the network)
+ // HiddenLayers <string> Layout of the hidden layers (nodes per layer)
+ // * specifiactions for each hidden layer are separated by commata
+ // * for each layer the number of nodes can be either absolut (simply a number)
+ // or relative to the number of input nodes to the neural net (N)
+ // * there is always a single node in the output layer
+ // example: a net with 6 input nodes and "Hiddenlayers=N-1,N-2" has 6,5,4,1 nodes in the
+ // layers 1,2,3,4, repectively
DeclareOptionRef(fNcycles=3000,"NCycles","Number of training cycles");
DeclareOptionRef(fLayerSpec="N-1,N-2","HiddenLayers","Specification of the hidden layers");
}
@@ -164,6 +174,8 @@ void TMVA::MethodTMlpANN::DeclareOptions()
//_______________________________________________________________________
void TMVA::MethodTMlpANN::ProcessOptions()
{
+ // builds the neural network as specified by the user
+
CreateMLPOptions(fLayerSpec);
// Here we create a dummy tree necessary to create
@@ -188,6 +200,7 @@ void TMVA::MethodTMlpANN::ProcessOptions()
//_______________________________________________________________________
Double_t TMVA::MethodTMlpANN::GetMvaValue()
{
+ // calculate the value of the neural net for the current event
static Double_t* d = new Double_t[Data().GetNVariables()];
for(UInt_t ivar = 0; ivar<Data().GetNVariables(); ivar++) {
d[ivar] = (Double_t)Data().Event().GetVal(ivar);
diff --git a/tmva/src/MethodVariable.cxx b/tmva/src/MethodVariable.cxx
index 67012064749..25378e68c34 100644
--- a/tmva/src/MethodVariable.cxx
+++ b/tmva/src/MethodVariable.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MethodVariable.cxx,v 1.21 2006/11/02 15:44:50 andreas.hoecker Exp $
+// @(#)root/tmva $Id: MethodVariable.cxx,v 1.24 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -50,7 +50,7 @@ TMVA::MethodVariable::MethodVariable( TString jobName, TString methodTitle, Data
// so, remove this part
SetMethodName( "Variable" );
- SetMethodType( TMVA::Types::Variable );
+ SetMethodType( TMVA::Types::kVariable );
SetTestvarPrefix( "" );
SetTestvarName();
@@ -93,11 +93,13 @@ Double_t TMVA::MethodVariable::GetMvaValue()
//_______________________________________________________________________
void TMVA::MethodVariable::WriteWeightsToStream( ostream & o ) const
{
+ // Write configuration to output stream
o << "";
}
//_______________________________________________________________________
void TMVA::MethodVariable::ReadWeightsFromStream( istream & istr )
{
+ // Read configuration from input stream
if (istr.eof());
}
diff --git a/tmva/src/MisClassificationError.cxx b/tmva/src/MisClassificationError.cxx
index a07bdbee008..5747d6e41d5 100644
--- a/tmva/src/MisClassificationError.cxx
+++ b/tmva/src/MisClassificationError.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MisClassificationError.cxx,v 1.11 2006/11/06 00:10:17 helgevoss Exp $
+// @(#)root/tmva $Id: MisClassificationError.cxx,v 1.12 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,7 +13,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/src/MsgLogger.cxx b/tmva/src/MsgLogger.cxx
index 7a5fa49ba95..29fa62fda72 100644
--- a/tmva/src/MsgLogger.cxx
+++ b/tmva/src/MsgLogger.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: MsgLogger.cxx,v 1.6 2006/11/13 23:43:34 stelzer Exp $
+// @(#)root/tmva $Id: MsgLogger.cxx,v 1.10 2006/11/16 22:51:59 helgevoss Exp $
// Author: Attila Krasznahorkay
/**********************************************************************************
@@ -15,7 +15,7 @@
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -48,7 +48,7 @@ static const string PREFIX = "--- ";
// this is the hardcoded suffix
static const string SUFFIX = ": ";
-TMVA::MsgLogger::MsgLogger( const TObject* source, MsgType minType )
+TMVA::MsgLogger::MsgLogger( const TObject* source, EMsgType minType )
: fObjSource( source ),
fStrSource( "" ),
fPrefix( PREFIX ),
@@ -61,7 +61,7 @@ TMVA::MsgLogger::MsgLogger( const TObject* source, MsgType minType )
InitMaps();
}
-TMVA::MsgLogger::MsgLogger( const string& source, MsgType minType )
+TMVA::MsgLogger::MsgLogger( const string& source, EMsgType minType )
: fObjSource( 0 ),
fStrSource( source ),
fPrefix( PREFIX ),
@@ -74,7 +74,7 @@ TMVA::MsgLogger::MsgLogger( const string& source, MsgType minType )
InitMaps();
}
-TMVA::MsgLogger::MsgLogger( MsgType minType )
+TMVA::MsgLogger::MsgLogger( EMsgType minType )
: fObjSource( 0 ),
fStrSource( "Unknown" ),
fPrefix( PREFIX ),
@@ -96,19 +96,25 @@ TMVA::MsgLogger::MsgLogger( const MsgLogger& parent ) :
fSuffix( SUFFIX ),
fMaxSourceSize( MAXIMUM_SOURCE_NAME_LENGTH )
{
+ // copy constructor
InitMaps();
*this = parent;
}
TMVA::MsgLogger::~MsgLogger()
-{}
+{
+ // destructor
+}
TMVA::MsgLogger& TMVA::MsgLogger::operator= ( const MsgLogger& parent )
{
- fObjSource = parent.fObjSource;
- fStrSource = parent.fStrSource;
- fActiveType = parent.fActiveType;
- fMinType = parent.fMinType;
+ // assingment operator
+ if( &parent != this) {
+ fObjSource = parent.fObjSource;
+ fStrSource = parent.fStrSource;
+ fActiveType = parent.fActiveType;
+ fMinType = parent.fMinType;
+ }
return *this;
}
@@ -169,15 +175,19 @@ void TMVA::MsgLogger::Send()
return;
}
-void TMVA::MsgLogger::WriteMsg( MsgType type, const std::string& line ) const
+void TMVA::MsgLogger::WriteMsg( EMsgType type, const std::string& line ) const
{
+ // putting the output string, the message type, and the color
+ // switcher together into a single string
+
if (type < fMinType) return;
- map<MsgType, std::string>::const_iterator stype;
+ map<EMsgType, std::string>::const_iterator stype;
if ((stype = fTypeMap.find( type )) == fTypeMap.end()) return;
#ifdef USE_COLORED_CONSOLE
// no text for INFO
if (type == kINFO)
- cout << fColorMap.find( type )->second << fPrefix << line << "\033[0m" << endl;
+ // no color for info
+ cout << fPrefix << line << endl;
else
cout << fColorMap.find( type )->second << fPrefix << "<" << stype->second << "> " << line << "\033[0m" << endl;
#else
@@ -191,7 +201,7 @@ void TMVA::MsgLogger::WriteMsg( MsgType type, const std::string& line ) const
if (type == kFATAL) { cout << "***> abort program execution" << endl; exit(1); }
}
-TMVA::MsgLogger& TMVA::MsgLogger::endmsg( MsgLogger& logger )
+TMVA::MsgLogger& TMVA::MsgLogger::Endmsg( MsgLogger& logger )
{
// end line
logger.Send();
@@ -200,6 +210,7 @@ TMVA::MsgLogger& TMVA::MsgLogger::endmsg( MsgLogger& logger )
void TMVA::MsgLogger::InitMaps()
{
+ // fill maps that assign a string and a color to echo message level
fTypeMap[kVERBOSE] = "VERBOSE";
fTypeMap[kDEBUG] = "DEBUG";
fTypeMap[kINFO] = "INFO";
diff --git a/tmva/src/Node.cxx b/tmva/src/Node.cxx
index e448d99da4d..b52b92c4b82 100644
--- a/tmva/src/Node.cxx
+++ b/tmva/src/Node.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Node.cxx,v 1.20 2006/11/13 15:49:49 helgevoss Exp $
+// @(#)root/tmva $Id: Node.cxx,v 1.21 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* CopyRight (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/Option.cxx b/tmva/src/Option.cxx
index 927fc273c4d..17d2aecdc74 100644
--- a/tmva/src/Option.cxx
+++ b/tmva/src/Option.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Option.cxx,v 1.14 2006/11/14 13:56:12 stelzer Exp $
+// @(#)root/tmva $Id: Option.cxx,v 1.15 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/PDF.cxx b/tmva/src/PDF.cxx
index 0f29b9c5c7d..bfbd43d6ae7 100644
--- a/tmva/src/PDF.cxx
+++ b/tmva/src/PDF.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: PDF.cxx,v 1.20 2006/10/15 12:06:32 andreas.hoecker Exp $
+// @(#)root/tmva $Id: PDF.cxx,v 1.22 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -46,7 +46,7 @@ ClassImp(TMVA::PDF)
;
//_______________________________________________________________________
-TMVA::PDF::PDF( const TH1 *hist, TMVA::PDF::SmoothMethod method, Int_t nsmooth )
+TMVA::PDF::PDF( const TH1 *hist, TMVA::PDF::ESmoothMethod method, Int_t nsmooth )
: fUseHistogram( kFALSE ),
fNsmooth ( nsmooth ),
fSpline ( 0 ),
diff --git a/tmva/src/Ranking.cxx b/tmva/src/Ranking.cxx
index 80337bb8e35..498327cb26a 100644
--- a/tmva/src/Ranking.cxx
+++ b/tmva/src/Ranking.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Ranking.cxx,v 1.10 2006/10/17 14:02:14 krasznaa Exp $
+// @(#)root/tmva $Id: Ranking.cxx,v 1.12 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,11 +13,11 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -39,6 +39,7 @@ ClassImp(TMVA::Ranking)
TMVA::Ranking::Ranking()
: fLogger( "", kINFO )
{
+ // default constructor
fRanking.clear();
}
@@ -47,24 +48,26 @@ TMVA::Ranking::Ranking( const TString& context, const TString& rankingDiscrimina
fRankingDiscriminatorName( rankingDiscriminatorName ),
fLogger( context.Data(), kINFO )
{
+ // constructor
fRanking.clear();
}
TMVA::Ranking::~Ranking()
{
+ // destructor
fRanking.clear();
}
void TMVA::Ranking::AddRank( Rank& rank )
{
- fRanking.push_back( rank );
-
+ // Add a new rank
// sort according to rank value (descending)
// Who the hell knows why this does not compile on windos.. write the sorting
// reversing myself... (means sorting in "descending" order)
// --> std::sort ( fRanking.begin(), fRanking.end() );
// --> std::reverse( fRanking.begin(), fRanking.end() );
-
+ fRanking.push_back( rank );
+
UInt_t sizeofarray=fRanking.size();
Rank temp(fRanking[0]);
for (unsigned int i=0; i<sizeofarray; i++) {
@@ -108,18 +111,24 @@ TMVA::Rank::Rank( TString variable, Double_t rankValue )
: fVariable( variable ),
fRankValue( rankValue ),
fRank( -1 )
-{}
+{
+ // constructor
+}
TMVA::Rank::~Rank()
-{}
+{
+ // destructor
+}
-Bool_t TMVA::Rank::operator < ( const Rank& other ) const
+Bool_t TMVA::Rank::operator< ( const Rank& other ) const
{
+ // comparison operator <
if (fRankValue < other.fRankValue) return true;
else return false;
}
-Bool_t TMVA::Rank::operator > ( const Rank& other ) const
+Bool_t TMVA::Rank::operator> ( const Rank& other ) const
{
+ // comparison operator >
if (fRankValue > other.fRankValue) return true;
else return false;
}
diff --git a/tmva/src/Reader.cxx b/tmva/src/Reader.cxx
index a46c8378566..761be48164b 100644
--- a/tmva/src/Reader.cxx
+++ b/tmva/src/Reader.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Reader.cxx,v 1.31 2006/11/13 23:43:34 stelzer Exp $
+// @(#)root/tmva $Id: Reader.cxx,v 1.35 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -14,13 +14,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -182,12 +182,14 @@ void TMVA::Reader::Init( void )
//_______________________________________________________________________
void TMVA::Reader::AddVariable( const TString& expression, float* datalink)
{
+ // Add a float variable or expression to the reader
Data().AddVariable(expression, 'F', (void*)datalink);
}
//_______________________________________________________________________
void TMVA::Reader::AddVariable( const TString& expression, int* datalink)
{
+ // Add a integer variable or expression to the reader
Data().AddVariable(expression, 'I', (void*)datalink);
}
@@ -221,53 +223,53 @@ TMVA::IMethod* TMVA::Reader::BookMVA( TString methodName, TString weightfile )
//_______________________________________________________________________
-TMVA::IMethod* TMVA::Reader::BookMVA( TMVA::Types::MVA methodType, TString weightfile )
+TMVA::IMethod* TMVA::Reader::BookMVA( TMVA::Types::EMVA methodType, TString weightfile )
{
- IMethod* method = 0;
// books MVA method from weightfile
+ IMethod* method = 0;
switch (methodType) {
- case (TMVA::Types::Cuts):
+ case (TMVA::Types::kCuts):
method = new TMVA::MethodCuts( Data(), weightfile );
break;
- case (TMVA::Types::Likelihood):
+ case (TMVA::Types::kLikelihood):
method = new TMVA::MethodLikelihood( Data(), weightfile );
break;
- case (TMVA::Types::PDERS):
+ case (TMVA::Types::kPDERS):
method = new TMVA::MethodPDERS( Data(), weightfile );
break;
- case (TMVA::Types::HMatrix):
+ case (TMVA::Types::kHMatrix):
method = new TMVA::MethodHMatrix( Data(), weightfile );
break;
- case (TMVA::Types::Fisher):
+ case (TMVA::Types::kFisher):
method = new TMVA::MethodFisher( Data(), weightfile );
break;
- case (TMVA::Types::CFMlpANN):
+ case (TMVA::Types::kCFMlpANN):
method = new TMVA::MethodCFMlpANN( Data(), weightfile );
break;
- case (TMVA::Types::TMlpANN):
+ case (TMVA::Types::kTMlpANN):
method = new TMVA::MethodTMlpANN( Data(), weightfile );
break;
- case (TMVA::Types::BDT):
+ case (TMVA::Types::kBDT):
method = new TMVA::MethodBDT( Data(), weightfile );
break;
- case (TMVA::Types::MLP):
+ case (TMVA::Types::kMLP):
method = new TMVA::MethodMLP( Data(), weightfile );
break;
- case (TMVA::Types::RuleFit):
+ case (TMVA::Types::kRuleFit):
method = new TMVA::MethodRuleFit( Data(), weightfile );
break;
- case (TMVA::Types::BayesClassifier):
+ case (TMVA::Types::kBayesClassifier):
method = new TMVA::MethodBayesClassifier( Data(), weightfile );
break;
@@ -287,6 +289,9 @@ TMVA::IMethod* TMVA::Reader::BookMVA( TMVA::Types::MVA methodType, TString weigh
//_______________________________________________________________________
Double_t TMVA::Reader::EvaluateMVA( const std::vector<Float_t>& inputVec, TString methodName, Double_t aux )
{
+ // Evaluate a vector<float> of input data for a given method
+ // The parameter aux is obligatory for the cuts method where it represents the efficiency cutoff
+
for (UInt_t ivar=0; ivar<inputVec.size(); ivar++) Data().Event().SetVal( ivar, inputVec[ivar] );
return EvaluateMVA( methodName, aux );
@@ -295,6 +300,9 @@ Double_t TMVA::Reader::EvaluateMVA( const std::vector<Float_t>& inputVec, TStrin
//_______________________________________________________________________
Double_t TMVA::Reader::EvaluateMVA( const std::vector<Double_t>& inputVec, TString methodName, Double_t aux )
{
+ // Evaluate a vector<double> of input data for a given method
+ // The parameter aux is obligatory for the cuts method where it represents the efficiency cutoff
+
for (UInt_t ivar=0; ivar<inputVec.size(); ivar++) Data().Event().SetVal( ivar, (Float_t)inputVec[ivar] );
return EvaluateMVA( methodName, aux );
@@ -303,9 +311,9 @@ Double_t TMVA::Reader::EvaluateMVA( const std::vector<Double_t>& inputVec, TStri
//_______________________________________________________________________
Double_t TMVA::Reader::EvaluateMVA( TString methodName, Double_t aux )
{
+ // evaluates MVA for given set of input variables
IMethod* method = 0;
- // evaluates MVA for given set of input variables
std::map<TString, IMethod*>::iterator it = fMethodMap.find( methodName );
if (it == fMethodMap.end()) {
for (it = fMethodMap.begin(); it!=fMethodMap.end(); it++) fLogger << "M" << it->first << Endl;
@@ -327,11 +335,11 @@ Double_t TMVA::Reader::EvaluateMVA( IMethod* method, Double_t aux )
// NOTE: in likelihood the preprocessing transformations are inserted by hand in GetMvaValue()
// (to distinguish signal and background transformations), and hence should not be applied here
- if (method->GetMethodType() != Types::Likelihood)
+ if (method->GetMethodType() != Types::kLikelihood)
Data().ApplyTransformation( method->GetPreprocessingMethod(), kTRUE );
// the aux value is only needed for MethodCuts: it sets the required signal efficiency
- if (method->GetMethodType() == TMVA::Types::Cuts)
+ if (method->GetMethodType() == TMVA::Types::kCuts)
((TMVA::MethodCuts*)method)->SetTestSignalEfficiency( aux );
Double_t mvaVal = method->GetMvaValue();
diff --git a/tmva/src/RootFinder.cxx b/tmva/src/RootFinder.cxx
index 0d3d9705b50..e2af19eb707 100644
--- a/tmva/src/RootFinder.cxx
+++ b/tmva/src/RootFinder.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: RootFinder.cxx,v 1.11 2006/10/15 22:34:22 andreas.hoecker Exp $
+// @(#)root/tmva $Id: RootFinder.cxx,v 1.12 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/Rule.cxx b/tmva/src/Rule.cxx
index 1188cfbea33..abdedb36149 100644
--- a/tmva/src/Rule.cxx
+++ b/tmva/src/Rule.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Rule.cxx,v 1.30 2006/10/29 23:40:57 helgevoss Exp $
+// @(#)root/tmva $Id: Rule.cxx,v 1.32 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -20,7 +20,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -56,7 +56,14 @@
TMVA::Rule::Rule( RuleEnsemble *re,
const std::vector< const Node * >& nodes,
const std::vector< Int_t > & cutdirs )
- : fLogger( "Rule" )
+ : fNorm ( 1.0 )
+ , fSupport ( 0.0 )
+ , fSigma ( 0.0 )
+ , fCoefficient ( 0.0 )
+ , fImportance ( 0.0 )
+ , fImportanceRef ( 1.0 )
+ , fRuleEnsemble ( re )
+ , fLogger( "Rule" )
{
// the main constructor for a Rule
@@ -65,15 +72,8 @@ TMVA::Rule::Rule( RuleEnsemble *re,
// nodes - a vector of TMVA::Node; from these all possible rules will be created
//
//
- fRuleEnsemble = re;
SetNodes( nodes );
SetCutDirs( cutdirs );
- fNorm = 1.0;
- fCoefficient = 0.0;
- fSupport = 0.0;
- fSigma = 0.0;
- fImportance = 0.0;
- fImportanceRef = 1.0;
}
//_______________________________________________________________________
@@ -274,18 +274,18 @@ Double_t TMVA::Rule::RuleDist( const Rule& other, Bool_t useCutValue ) const
//_______________________________________________________________________
void TMVA::Rule::GetEffectiveRule( std::vector<Int_t>& nodeind ) const
-//
-// Returns a vector of node indecis which correspond to the effective rule.
-// E.g, the rule:
-// v1<0.1
-// v1<0.05
-// v4>0.12
-//
-// is effectively the same as:
-// v1<0.05
-// v4>0.12
-//
{
+ //
+ // Returns a vector of node indecis which correspond to the effective rule.
+ // E.g, the rule:
+ // v1<0.1
+ // v1<0.05
+ // v4>0.12
+ //
+ // is effectively the same as:
+ // v1<0.05
+ // v4>0.12
+ //
nodeind.clear();
UInt_t nnodes = fNodes.size();
if (nnodes==2) { // just one cut, return all nodes
@@ -370,18 +370,22 @@ Bool_t TMVA::Rule::IsSimpleRule() const
//_______________________________________________________________________
Bool_t TMVA::Rule::operator==( const TMVA::Rule& other ) const
{
+ // comparison operator ==
+
return this->Equal( other, kTRUE, 1e-3 );
}
//_______________________________________________________________________
Bool_t TMVA::Rule::operator<( const TMVA::Rule& other ) const
{
+ // comparison operator <
return (fImportance < other.GetImportance());
}
//_______________________________________________________________________
ostream& TMVA::operator<< ( ostream& os, const TMVA::Rule& rule )
{
+ // ostream operator
rule.Print( os );
return os;
}
@@ -389,28 +393,34 @@ ostream& TMVA::operator<< ( ostream& os, const TMVA::Rule& rule )
//_______________________________________________________________________
const TString & TMVA::Rule::GetVarName( Int_t i ) const
{
+ // returns the name of a rule
+
return fRuleEnsemble->GetMethodRuleFit()->GetInputExp(i);
}
//_______________________________________________________________________
void TMVA::Rule::Copy( const Rule& other )
{
- SetRuleEnsemble( other.GetRuleEnsemble() );
- SetNodes( other.GetNodes() );
- fSSB = other.GetSSB();
- fSSBNeve = other.GetSSBNeve();
- SetCutDirs( other.GetCutDirs() );
- SetCoefficient(other.GetCoefficient());
- SetSupport( other.GetSupport() );
- SetSigma( other.GetSigma() );
- SetNorm( other.GetNorm() );
- CalcImportance();
- SetImportanceRef( other.GetImportanceRef() );
+ // copy function
+ if(this != &other) {
+ SetRuleEnsemble( other.GetRuleEnsemble() );
+ SetNodes( other.GetNodes() );
+ fSSB = other.GetSSB();
+ fSSBNeve = other.GetSSBNeve();
+ SetCutDirs( other.GetCutDirs() );
+ SetCoefficient(other.GetCoefficient());
+ SetSupport( other.GetSupport() );
+ SetSigma( other.GetSigma() );
+ SetNorm( other.GetNorm() );
+ CalcImportance();
+ SetImportanceRef( other.GetImportanceRef() );
+ }
}
//_______________________________________________________________________
void TMVA::Rule::Print( ostream& os ) const
{
+ // print function
Int_t ind;
Int_t sel,ntype,nnodes;
Double_t data, ssbval;
@@ -449,6 +459,7 @@ void TMVA::Rule::Print( ostream& os ) const
//_______________________________________________________________________
void TMVA::Rule::PrintRaw( ostream& os ) const
{
+ // extensive print function
const TMVA::DecisionTreeNode *node;
std::vector<Int_t> nodes;
GetEffectiveRule( nodes );
@@ -484,6 +495,8 @@ void TMVA::Rule::PrintRaw( ostream& os ) const
//_______________________________________________________________________
void TMVA::Rule::ReadRaw( istream& istr )
{
+ // read function (format is the same as written by PrintRaw)
+
TString dummy;
TMVA::DecisionTreeNode *node;
std::vector<Int_t> nodes;
diff --git a/tmva/src/RuleEnsemble.cxx b/tmva/src/RuleEnsemble.cxx
index 20647acc17f..7b5bea09e6c 100644
--- a/tmva/src/RuleEnsemble.cxx
+++ b/tmva/src/RuleEnsemble.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: RuleEnsemble.cxx,v 1.30 2006/11/14 15:39:36 helgevoss Exp $
+// @(#)root/tmva $Id: RuleEnsemble.cxx,v 1.33 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -19,7 +19,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -38,12 +38,12 @@
//_______________________________________________________________________
TMVA::RuleEnsemble::RuleEnsemble( RuleFit *rf )
- : fLogger( "RuleEnsemble" )
+ : fLearningModel ( kFull )
+ , fAverageRuleSigma ( 0.4 ) // default value - used if only linear model is chosen
+ , fMaxRuleDist ( 1e-3 ) // closest allowed 'distance' between two rules
+ , fLogger( "RuleEnsemble" )
{
// constructor
- fAverageRuleSigma = 0.4; // default value - used if only linear model is chosen
- fLearningModel = kFull;
- fMaxRuleDist = 1e-3; // closest allowed 'distance' between two rules
Initialize( rf );
}
@@ -51,7 +51,7 @@ TMVA::RuleEnsemble::RuleEnsemble( RuleFit *rf )
TMVA::RuleEnsemble::RuleEnsemble( const RuleEnsemble& other )
: fLogger( "RuleEnsemble" )
{
- // constructor
+ // copy constructor
Copy( other );
}
@@ -65,6 +65,7 @@ TMVA::RuleEnsemble::RuleEnsemble()
//_______________________________________________________________________
TMVA::RuleEnsemble::~RuleEnsemble()
{
+ // destructor
for ( std::vector< TMVA::Rule *>::iterator itrRule = fRules.begin(); itrRule != fRules.end(); itrRule++ ) {
delete *itrRule;
}
@@ -74,6 +75,8 @@ TMVA::RuleEnsemble::~RuleEnsemble()
//_______________________________________________________________________
void TMVA::RuleEnsemble::Initialize( RuleFit *rf )
{
+ // Initializes all member variables with default values
+
fAverageRuleSigma = 0.4; // default value - used if only linear model is chosen
fRuleFit = rf;
UInt_t nvars = GetMethodRuleFit()->GetNvar();
@@ -98,6 +101,7 @@ const TMVA::MethodRuleFit* TMVA::RuleEnsemble::GetMethodRuleFit() const
//_______________________________________________________________________
void TMVA::RuleEnsemble::MakeModel()
{
+ // create model
if (DoRules())
MakeRules( fRuleFit->GetForest() );
if (DoLinear())
@@ -125,6 +129,8 @@ Double_t TMVA::RuleEnsemble::CoefficientRadius()
//_______________________________________________________________________
void TMVA::RuleEnsemble::ResetCoefficients()
{
+ // reset all rule coefficients
+
fOffset = 0.0;
UInt_t nrules = fRules.size();
for (UInt_t i=0; i<nrules; i++) {
@@ -135,6 +141,8 @@ void TMVA::RuleEnsemble::ResetCoefficients()
//_______________________________________________________________________
void TMVA::RuleEnsemble::SetCoefficients( const std::vector< Double_t > & v )
{
+ // set all rule coefficients
+
UInt_t nrules = fRules.size();
if (v.size()!=nrules) {
fLogger << kFATAL << "<SetCoefficients> - BUG TRAP - input vector worng size! It is = " << v.size()
@@ -148,6 +156,8 @@ void TMVA::RuleEnsemble::SetCoefficients( const std::vector< Double_t > & v )
//_______________________________________________________________________
void TMVA::RuleEnsemble::GetCoefficients( std::vector< Double_t > & v )
{
+ // Retrieve all rule coefficients
+
UInt_t nrules = fRules.size();
v.resize(nrules);
if (nrules==0) return;
@@ -160,41 +170,50 @@ void TMVA::RuleEnsemble::GetCoefficients( std::vector< Double_t > & v )
//_______________________________________________________________________
const std::vector<const TMVA::Event *> *TMVA::RuleEnsemble::GetTrainingEvents() const
{
+ // get list of training events from the rule fitter
+
return &(fRuleFit->GetTrainingEvents());
}
//_______________________________________________________________________
const std::vector< Int_t > *TMVA::RuleEnsemble::GetSubsampleEvents() const
{
+ // get list of events for the subsamples from the rule fitter
return &(fRuleFit->GetSubsampleEvents());
}
//_______________________________________________________________________
void TMVA::RuleEnsemble::GetSubsampleEvents(UInt_t sub, UInt_t & ibeg, UInt_t & iend) const
{
+ // get list of events for the subsample sub from the rule fitter
fRuleFit->GetSubsampleEvents(sub,ibeg,iend);
}
//_______________________________________________________________________
const UInt_t TMVA::RuleEnsemble::GetNSubsamples() const
{
+ // get the number of subsamples from the rule fitter
return fRuleFit->GetNSubsamples();
}
//_______________________________________________________________________
const TMVA::Event * TMVA::RuleEnsemble::GetTrainingEvent(UInt_t i) const
{
+ // get the training event from the rule fitter
return fRuleFit->GetTrainingEvent(i);
}
//_______________________________________________________________________
const TMVA::Event * TMVA::RuleEnsemble::GetTrainingEvent(UInt_t i, UInt_t isub) const
{
+ // get one training event for one subsample from the rule fitter
return fRuleFit->GetTrainingEvent(i,isub);
}
//_______________________________________________________________________
void TMVA::RuleEnsemble::SetRulesNCuts()
{
+ // set the number of nodes to the cut array
+
std::vector<Int_t> nodes;
fRulesNCuts.clear();
for (UInt_t i=0; i<fRules.size(); i++) {
@@ -206,6 +225,8 @@ void TMVA::RuleEnsemble::SetRulesNCuts()
//_______________________________________________________________________
void TMVA::RuleEnsemble::RemoveSimpleRules()
{
+ // remove all simple rules
+
fLogger << kINFO << "removing simple rules" << Endl;
UInt_t nrulesIn = fRules.size();
std::vector<bool> removeMe( nrulesIn,false );
@@ -234,6 +255,8 @@ void TMVA::RuleEnsemble::RemoveSimpleRules()
//_______________________________________________________________________
void TMVA::RuleEnsemble::RemoveSimilarRules()
{
+ // remove rules that behave similar
+
fLogger << kINFO << "removing similar rules; distance = " << fMaxRuleDist << Endl;
UInt_t nrulesIn = fRules.size();
@@ -287,6 +310,8 @@ void TMVA::RuleEnsemble::RemoveSimilarRules()
//_______________________________________________________________________
void TMVA::RuleEnsemble::CleanupRules()
{
+ // cleanup rules
+
UInt_t nrules = fRules.size();
if (nrules==0) return;
fLogger << kINFO << "removing rules with relative importance < " << fImportanceCut << Endl;
@@ -312,6 +337,8 @@ void TMVA::RuleEnsemble::CleanupRules()
//_______________________________________________________________________
void TMVA::RuleEnsemble::CleanupLinear()
{
+ // cleanup linear model
+
UInt_t nlin = fLinNorm.size();
if (nlin==0) return;
fLogger << kINFO << "removing linear terms with relative importance < " << fImportanceCut << Endl;
@@ -326,6 +353,8 @@ void TMVA::RuleEnsemble::CleanupLinear()
//_______________________________________________________________________
void TMVA::RuleEnsemble::CalcRuleSupport()
{
+ // calculate the support for all rules
+
Double_t seve, s,t,stot,ttot;
Double_t ssig, sbkg;
Int_t indrule=0;
@@ -371,6 +400,8 @@ void TMVA::RuleEnsemble::CalcRuleSupport()
//_______________________________________________________________________
void TMVA::RuleEnsemble::CalcImportance()
{
+ // calculate the importance of each rule
+
Double_t maxRuleImp = CalcRuleImportance();
Double_t maxLinImp = CalcLinImportance();
Double_t maxImp = (maxRuleImp>maxLinImp ? maxRuleImp : maxLinImp);
@@ -384,6 +415,8 @@ void TMVA::RuleEnsemble::CalcImportance()
//_______________________________________________________________________
Double_t TMVA::RuleEnsemble::CalcRuleImportance()
{
+ // calculate importance of each rule
+
Double_t maxImp=-1.0;
Double_t imp;
Int_t nrules = fRules.size();
@@ -402,6 +435,8 @@ Double_t TMVA::RuleEnsemble::CalcRuleImportance()
//_______________________________________________________________________
Double_t TMVA::RuleEnsemble::CalcLinImportance()
{
+ // calculate the linear importance for each rule
+
Double_t maxImp=-1.0;
UInt_t nvars = fLinCoefficients.size();
fLinImportance.resize(nvars,0.0);
@@ -661,7 +696,8 @@ Double_t TMVA::RuleEnsemble::FStar() const
//_____________________________________________________________________
Double_t TMVA::RuleEnsemble::EvalEvent() const
{
- //
+ // evaluate current event
+
Int_t nrules = fRules.size();
Double_t rval=fOffset;
Double_t linear=0;
@@ -686,6 +722,7 @@ Double_t TMVA::RuleEnsemble::EvalEvent() const
//_____________________________________________________________________
Double_t TMVA::RuleEnsemble::EvalEvent(const TMVA::Event & e)
{
+ // evaluate event e
SetEvent(e);
UpdateEventVal();
return EvalEvent();
@@ -694,6 +731,8 @@ Double_t TMVA::RuleEnsemble::EvalEvent(const TMVA::Event & e)
//_______________________________________________________________________
Double_t TMVA::RuleEnsemble::EvalLinEventRaw( UInt_t vind, const TMVA::Event & e)
{
+ // evaluate the event linearly (not normalized)
+
Double_t val = e.GetVal(vind);
Double_t rval = TMath::Min( fLinDP[vind], TMath::Max( fLinDM[vind], val ) );
return rval;
@@ -702,6 +741,7 @@ Double_t TMVA::RuleEnsemble::EvalLinEventRaw( UInt_t vind, const TMVA::Event & e
//_______________________________________________________________________
Double_t TMVA::RuleEnsemble::EvalLinEvent( UInt_t vind, Bool_t norm ) const
{
+ // evaluate the event linearly normalized
Double_t rval=0;
rval = fEventLinearVal[vind];
if (norm) rval*=fLinNorm[vind];
@@ -711,6 +751,8 @@ Double_t TMVA::RuleEnsemble::EvalLinEvent( UInt_t vind, Bool_t norm ) const
//_______________________________________________________________________
Double_t TMVA::RuleEnsemble::EvalLinEvent() const
{
+ // evaluate event linearly
+
Double_t rval=0;
for (UInt_t v=0; v<fLinTermOK.size(); v++) {
if (fLinTermOK[v])
@@ -722,6 +764,8 @@ Double_t TMVA::RuleEnsemble::EvalLinEvent() const
//_______________________________________________________________________
Double_t TMVA::RuleEnsemble::EvalLinEvent( const TMVA::Event& e )
{
+ // evaluate event linearly
+
SetEvent(e);
UpdateEventVal();
return EvalLinEvent();
@@ -730,6 +774,8 @@ Double_t TMVA::RuleEnsemble::EvalLinEvent( const TMVA::Event& e )
//_______________________________________________________________________
void TMVA::RuleEnsemble::RuleStatistics()
{
+ // calculate various statistics for this rule
+
const UInt_t nrules = fRules.size();
const std::vector<const TMVA::Event *> *events = GetTrainingEvents();
const TMVA::Event *eveData;
@@ -802,6 +848,8 @@ void TMVA::RuleEnsemble::RuleStatistics()
//_______________________________________________________________________
void TMVA::RuleEnsemble::Print( ostream& os ) const
{
+ // print function
+
fLogger << kINFO << "===============================" << Endl;
fLogger << kINFO << " RuleEnsemble " << Endl;
fLogger << kINFO << "===============================" << Endl;
@@ -961,18 +1009,22 @@ void TMVA::RuleEnsemble::ReadRaw( istream & istr )
//_______________________________________________________________________
void TMVA::RuleEnsemble::Copy( const RuleEnsemble & other )
{
- fRuleFit = other.GetRuleFit();
- fMaxRuleDist = other.GetMaxRuleDist();
- fOffset = other.GetOffset();
- fRules = other.GetRulesConst();
- fImportanceCut = other.GetImportanceCut();
- fVarImportance = other.GetVarImportance();
- fLearningModel = other.GetLearningModel();
+ // copy function
+ if(this != &other) {
+ fRuleFit = other.GetRuleFit();
+ fMaxRuleDist = other.GetMaxRuleDist();
+ fOffset = other.GetOffset();
+ fRules = other.GetRulesConst();
+ fImportanceCut = other.GetImportanceCut();
+ fVarImportance = other.GetVarImportance();
+ fLearningModel = other.GetLearningModel();
+ }
}
//_______________________________________________________________________
Int_t TMVA::RuleEnsemble::CalcNRules( const TMVA::DecisionTree *dtree )
{
+ // calculate the number of rules
if (dtree==0) return 0;
TMVA::Node *node = dtree->GetRoot();
Int_t nendnodes = 0;
@@ -983,6 +1035,8 @@ Int_t TMVA::RuleEnsemble::CalcNRules( const TMVA::DecisionTree *dtree )
//_______________________________________________________________________
void TMVA::RuleEnsemble::FindNEndNodes( const TMVA::Node *node, Int_t & nendnodes )
{
+ // find the number of leaf nodes
+
if (node==0) return;
if (dynamic_cast<const TMVA::DecisionTreeNode*>(node)->GetSelector()<0) {
++nendnodes;
@@ -997,6 +1051,7 @@ void TMVA::RuleEnsemble::FindNEndNodes( const TMVA::Node *node, Int_t & nendnode
//_______________________________________________________________________
void TMVA::RuleEnsemble::MakeRulesFromTree( const TMVA::DecisionTree *dtree )
{
+ // create rules from the decsision tree structure
TMVA::Node *node = dtree->GetRoot();
AddRule( node );
}
@@ -1004,6 +1059,8 @@ void TMVA::RuleEnsemble::MakeRulesFromTree( const TMVA::DecisionTree *dtree )
//_______________________________________________________________________
void TMVA::RuleEnsemble::AddRule( const TMVA::Node *node )
{
+ // add a new rule to the tree
+
if (node==0) return;
if (node->GetParent()==0) { // it's a root node, don't make a rule
AddRule( node->GetRight() );
@@ -1081,6 +1138,7 @@ TMVA::Rule *TMVA::RuleEnsemble::MakeTheRule( const TMVA::Node *node )
//_______________________________________________________________________
ostream& TMVA::operator<< ( ostream& os, const TMVA::RuleEnsemble & rules )
{
+ // ostream operator
rules.Print( os );
return os;
}
diff --git a/tmva/src/RuleFit.cxx b/tmva/src/RuleFit.cxx
index eebd656d8c0..0e939007a9e 100644
--- a/tmva/src/RuleFit.cxx
+++ b/tmva/src/RuleFit.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: RuleFit.cxx,v 1.26 2006/10/17 21:22:30 andreas.hoecker Exp $
+// @(#)root/tmva $Id: RuleFit.cxx,v 1.28 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -21,7 +21,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -44,17 +44,22 @@ TMVA::RuleFit::RuleFit( const TMVA::MethodRuleFit *rfbase,
Double_t samplefrac )
: fLogger( "RuleFit" )
{
+ // constructor
Initialise( rfbase, forest, trainingEvents, samplefrac );
}
//_______________________________________________________________________
TMVA::RuleFit::RuleFit()
: fLogger( "RuleFit" )
-{}
+{
+ // default constructor
+}
//_______________________________________________________________________
TMVA::RuleFit::~RuleFit()
-{}
+{
+ // destructor
+}
//_______________________________________________________________________
void TMVA::RuleFit::Initialise( const TMVA::MethodRuleFit *rfbase,
@@ -62,6 +67,7 @@ void TMVA::RuleFit::Initialise( const TMVA::MethodRuleFit *rfbase,
const std::vector< TMVA::Event *> & events,
Double_t sampfrac )
{
+ // initialize the parameters of the RuleFit method
fMethodRuleFit = rfbase;
std::vector< TMVA::DecisionTree *>::const_iterator itrDtree=forest.begin();
for (; itrDtree!=forest.end(); ++itrDtree ) fForest.push_back( *itrDtree );
@@ -84,19 +90,23 @@ void TMVA::RuleFit::Initialise( const TMVA::MethodRuleFit *rfbase,
//_______________________________________________________________________
void TMVA::RuleFit::Copy( const TMVA::RuleFit& other )
{
- fMethodRuleFit = other.GetMethodRuleFit();
- fTrainingEvents = other.GetTrainingEvents();
- fSubsampleEvents = other.GetSubsampleEvents();
+ // copy method
+ if(this != &other) {
+ fMethodRuleFit = other.GetMethodRuleFit();
+ fTrainingEvents = other.GetTrainingEvents();
+ fSubsampleEvents = other.GetSubsampleEvents();
- fForest = other.GetForest();
- fRuleEnsemble = other.GetRuleEnsemble();
+ fForest = other.GetForest();
+ fRuleEnsemble = other.GetRuleEnsemble();
+ }
}
//_______________________________________________________________________
void TMVA::RuleFit::ForestStatistics()
-// summary of statistics of all trees
-// * end-nodes: average and spread
{
+ // summary of statistics of all trees
+ // * end-nodes: average and spread
+
UInt_t ntrees = fForest.size();
Double_t nt = Double_t(ntrees);
const TMVA::DecisionTree *tree;
@@ -128,6 +138,8 @@ void TMVA::RuleFit::FitCoefficients()
//_______________________________________________________________________
void TMVA::RuleFit::CalcImportance()
{
+ // calculates the importance of each rule
+
fLogger << kINFO << "calculating importance" << Endl;
fRuleEnsemble.CalcImportance();
fRuleEnsemble.CleanupRules();
@@ -140,12 +152,16 @@ void TMVA::RuleFit::CalcImportance()
//_______________________________________________________________________
Double_t TMVA::RuleFit::EvalEvent( const TMVA::Event& e )
{
+ // evaluate single event
+
return fRuleEnsemble.EvalEvent( e );
}
//_______________________________________________________________________
void TMVA::RuleFit::SetTrainingEvents( const std::vector<TMVA::Event *>& el, Double_t sampfrac )
{
+ // set the training events randomly
+
UInt_t neve = el.size();
if (neve==0) fLogger << kWARNING << "an empty sample of training events was given" << Endl;
@@ -174,6 +190,8 @@ void TMVA::RuleFit::SetTrainingEvents( const std::vector<TMVA::Event *>& el, Dou
//_______________________________________________________________________
void TMVA::RuleFit::GetSubsampleEvents(Int_t sub, UInt_t& ibeg, UInt_t& iend) const
{
+ // get the events for subsample sub
+
Int_t nsub = GetNSubsamples();
if (nsub==0) {
fLogger << kFATAL << "<GetSubsampleEvents> - wrong size, not properly initialised! BUG!!!" << Endl;
diff --git a/tmva/src/RuleFitParams.cxx b/tmva/src/RuleFitParams.cxx
index 5c4c8fa46e7..9253afa4bd3 100644
--- a/tmva/src/RuleFitParams.cxx
+++ b/tmva/src/RuleFitParams.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: RuleFitParams.cxx,v 1.29 2006/11/14 15:39:36 helgevoss Exp $
+// @(#)root/tmva $Id: RuleFitParams.cxx,v 1.32 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
/**********************************************************************************
@@ -17,7 +17,7 @@
* Copyright (c) 2005: *
* CERN, Switzerland, *
* Iowa State U. *
- * MPI-KP Heidelberg, Germany *
+ * MPI-K Heidelberg, Germany *
* *
* Redistribution and use in source and binary forms, with or without *
* modification, are permitted according to the terms listed in LICENSE *
@@ -39,29 +39,29 @@
//_______________________________________________________________________
TMVA::RuleFitParams::RuleFitParams()
- : fLogger( "RuleFitParams" )
+ : fRuleFit ( 0 )
+ , fRuleEnsemble ( 0 )
+ , fPathIdx1 ( 0 )
+ , fPathIdx2 ( 0 )
+ , fPerfIdx1 ( 0 )
+ , fPerfIdx2 ( 0 )
+ , fGDTau ( 0.0 )
+ , fGDPathStep ( 0.01 )
+ , fGDNPathSteps ( 100 )
+ , fGDErrNsigma ( 1.0 )
+ , fGDNtuple ( 0 )
+ , fNTOffset ( 0 )
+ , fNTCoeff ( 0 )
+ , fNTLinCoeff ( 0 )
+ , fLogger( "RuleFitParams" )
{
- fRuleFit = 0;
- fRuleEnsemble = 0;
- fGDTau = 0.0;
- fGDPathStep = 0.01;
- fGDNPathSteps = 100;
- fGDNtuple = 0;
- fGDErrNsigma = 1.0;
- fNTLinCoeff = 0;
- fNTCoeff = 0;
- fNTOffset = 0;
-
- fPathIdx1 = 0;
- fPathIdx2 = 0;
- fPerfIdx1 = 0;
- fPerfIdx2 = 0;
-
+ // constructor
Init();
}
//_______________________________________________________________________
TMVA::RuleFitParams::~RuleFitParams()
{
+ // destructor
if (fNTCoeff) { delete fNTCoeff; fNTCoeff = 0; }
if (fNTLinCoeff) { delete fNTLinCoeff;fNTLinCoeff = 0; }
}
@@ -69,7 +69,7 @@ TMVA::RuleFitParams::~RuleFitParams()
//_______________________________________________________________________
void TMVA::RuleFitParams::Init()
{
- //
+ // Initializes all parameters using the RuleEnsemble and the training tree
if (fRuleFit==0) return;
fRuleEnsemble = fRuleFit->GetRuleEnsemblePtr();
UInt_t nrules = fRuleEnsemble->GetNRules();
@@ -102,6 +102,8 @@ void TMVA::RuleFitParams::Init()
//_______________________________________________________________________
void TMVA::RuleFitParams::InitNtuple()
{
+ // initializes the ntuple
+
const UInt_t nrules = fRuleEnsemble->GetNRules();
const UInt_t nlin = fRuleEnsemble->GetLinNorm().size();
//
@@ -132,24 +134,28 @@ void TMVA::RuleFitParams::InitNtuple()
//_______________________________________________________________________
const std::vector< Int_t > *TMVA::RuleFitParams::GetSubsampleEvents() const
{
+ // accessor to the subsamples
return &(fRuleFit->GetSubsampleEvents());
}
//_______________________________________________________________________
void TMVA::RuleFitParams::GetSubsampleEvents(UInt_t sub, UInt_t & ibeg, UInt_t & iend) const
{
+ // calls the Subsample Events
fRuleFit->GetSubsampleEvents(sub,ibeg,iend);
}
//_______________________________________________________________________
const UInt_t TMVA::RuleFitParams::GetNSubsamples() const
{
+ // get the number of subsamples
return fRuleFit->GetNSubsamples();
}
//_______________________________________________________________________
const TMVA::Event *TMVA::RuleFitParams::GetTrainingEvent(UInt_t i, UInt_t isub) const
{
+ // accesses a training event
return fRuleFit->GetTrainingEvent(i,isub);
}
@@ -158,8 +164,8 @@ Double_t TMVA::RuleFitParams::LossFunction( const TMVA::Event& e ) const
{
// Implementation of squared-error ramp loss function (eq 39,40 in ref 1)
// This is used for binary Classifications where y = {+1,-1} for (sig,bkg)
- Double_t H = max( -1.0, min(1.0,fRuleEnsemble->EvalEvent( e )) );
- Double_t diff = (e.IsSignal()?1:-1) - H;
+ Double_t h = max( -1.0, min(1.0,fRuleEnsemble->EvalEvent( e )) );
+ Double_t diff = (e.IsSignal()?1:-1) - h;
//
return diff*diff;
}
@@ -167,6 +173,7 @@ Double_t TMVA::RuleFitParams::LossFunction( const TMVA::Event& e ) const
//_______________________________________________________________________
Double_t TMVA::RuleFitParams::Risk(UInt_t ibeg, UInt_t iend) const
{
+ // risk asessment
UInt_t neve = iend-ibeg+1;
if (neve<1) {
fLogger << kWARNING << "makeGradientVector() - invalid start/end indices!" << Endl;
@@ -452,6 +459,8 @@ void TMVA::RuleFitParams::MakeGDPath()
//_______________________________________________________________________
void TMVA::RuleFitParams::FillCoefficients()
{
+ // helper function to store the rule coefficients in local arrays
+
const UInt_t nrules = fRuleEnsemble->GetNRules();
const UInt_t nlin = fRuleEnsemble->GetLinNorm().size();
//
@@ -468,7 +477,6 @@ void TMVA::RuleFitParams::FillCoefficients()
//_______________________________________________________________________
void TMVA::RuleFitParams::CalcFStar(UInt_t ibeg, UInt_t iend)
{
- //
// Estimates F* (optimum scoring function) for all events for the given sets.
// The result is used in ErrorRateReg().
//
@@ -828,6 +836,8 @@ void TMVA::RuleFitParams::UpdateCoefficients()
//_______________________________________________________________________
Double_t TMVA::RuleFitParams::CalcAverageResponse(UInt_t ibeg, UInt_t iend)
{
+ // calulate the average response
+
UInt_t neve = iend-ibeg+1;
if (neve<1) {
fLogger << kFATAL << "<CalcAverageResponse> invalid start/end indices!" << Endl;
diff --git a/tmva/src/SdivSqrtSplusB.cxx b/tmva/src/SdivSqrtSplusB.cxx
index 315558b2942..7b609e24b02 100644
--- a/tmva/src/SdivSqrtSplusB.cxx
+++ b/tmva/src/SdivSqrtSplusB.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: SdivSqrtSplusB.cxx,v 1.8 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: SdivSqrtSplusB.cxx,v 1.9 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -14,7 +14,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/src/SeparationBase.cxx b/tmva/src/SeparationBase.cxx
index 6fc317063b1..65064f90e2b 100644
--- a/tmva/src/SeparationBase.cxx
+++ b/tmva/src/SeparationBase.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: SeparationBase.cxx,v 1.11 2006/11/06 00:10:17 helgevoss Exp $
+// @(#)root/tmva $Id: SeparationBase.cxx,v 1.12 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -28,7 +28,7 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
diff --git a/tmva/src/SimulatedAnnealingBase.cxx b/tmva/src/SimulatedAnnealingBase.cxx
index afa16215d50..69fa327cb40 100644
--- a/tmva/src/SimulatedAnnealingBase.cxx
+++ b/tmva/src/SimulatedAnnealingBase.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: SimulatedAnnealingBase.cxx,v 1.8 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: SimulatedAnnealingBase.cxx,v 1.10 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -39,23 +39,24 @@ ClassImp(TMVA::SimulatedAnnealingBase)
;
TMVA::SimulatedAnnealingBase::SimulatedAnnealingBase( std::vector<LowHigh_t*>& ranges )
- : fRanges( ranges )
+ : fRandom ( new TRandom() )
+ , fRanges ( ranges )
+ , fMaxCalls ( 500000 )
+ , fTemperatureGradient ( 0.3 )
+ , fUseAdaptiveTemperature( kFALSE )
+ , fInitialTemperature ( 1000 )
+ , fMinTemperature ( 0 )
+ , fEps ( 1e-04 )
+ , fNFunLoops ( 25 )
+ , fNEps ( 4 ) // needs to be at leas 2 !
{
- fRandom = new TRandom();
-
- // set default options
- fMaxCalls = 500000;
- fTemperatureGradient = 0.3;
- fUseAdaptiveTemperature = kFALSE;
- fInitialTemperature = 1000;
- fMinTemperature = 0;
- fEps = 1e-04;
- fNFunLoops = 25;
- fNEps = 4; // needs to be at leas 2 !
+ // constructor
}
TMVA::SimulatedAnnealingBase::~SimulatedAnnealingBase()
-{}
+{
+ // destructor
+}
Double_t TMVA::SimulatedAnnealingBase::Minimize( std::vector<Double_t>& parameters )
{
@@ -242,5 +243,6 @@ Double_t TMVA::SimulatedAnnealingBase::Minimize( std::vector<Double_t>& paramete
Double_t TMVA::SimulatedAnnealingBase::GetPerturbationProbability( Double_t E, Double_t Eref,
Double_t temperature )
{
+ // calculates the probability that a perturbation occured
return (temperature > 0) ? TMath::Exp( (E - Eref)/temperature ) : 0;
}
diff --git a/tmva/src/SimulatedAnnealingCuts.cxx b/tmva/src/SimulatedAnnealingCuts.cxx
index 623819de9e9..78563913cad 100644
--- a/tmva/src/SimulatedAnnealingCuts.cxx
+++ b/tmva/src/SimulatedAnnealingCuts.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: SimulatedAnnealingCuts.cxx,v 1.5 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: SimulatedAnnealingCuts.cxx,v 1.7 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -41,13 +41,18 @@ ClassImp(TMVA::SimulatedAnnealingCuts)
TMVA::SimulatedAnnealingCuts::SimulatedAnnealingCuts( std::vector<LowHigh_t*>& ranges )
: SimulatedAnnealingBase( ranges )
-{}
+{
+ // constructor
+}
TMVA::SimulatedAnnealingCuts::~SimulatedAnnealingCuts()
-{}
+{
+ // destructor
+}
Double_t TMVA::SimulatedAnnealingCuts::MinimizeFunction( const std::vector<Double_t>& parameters )
{
// minimize function interface for Simulated Annealing fitter for cut optimisation
+
return TMVA::MethodCuts::ThisCuts()->ComputeEstimator( parameters );
}
diff --git a/tmva/src/TNeuron.cxx b/tmva/src/TNeuron.cxx
index ed5fe1d9f36..c2fcf7f19f0 100644
--- a/tmva/src/TNeuron.cxx
+++ b/tmva/src/TNeuron.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: TNeuron.cxx,v 1.18 2006/10/15 12:06:33 andreas.hoecker Exp $
+// @(#)root/tmva $Id: TNeuron.cxx,v 1.19 2006/11/16 19:42:44 stelzer Exp $
// Author: Matt Jachowski
/**********************************************************************************
@@ -326,7 +326,7 @@ void TMVA::TNeuron::PrintActivationEqn()
}
//______________________________________________________________________________
-void TMVA::TNeuron::PrintMessage( MsgType type, TString message)
+void TMVA::TNeuron::PrintMessage( EMsgType type, TString message)
{
// print message, for debugging
fLogger << type << message << Endl;
diff --git a/tmva/src/TSpline1.cxx b/tmva/src/TSpline1.cxx
index a77c7f92c4f..184ec8bef11 100644
--- a/tmva/src/TSpline1.cxx
+++ b/tmva/src/TSpline1.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: TSpline1.cxx,v 1.10 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: TSpline1.cxx,v 1.11 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/TSpline2.cxx b/tmva/src/TSpline2.cxx
index a0167424863..f517305cbf2 100644
--- a/tmva/src/TSpline2.cxx
+++ b/tmva/src/TSpline2.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: TSpline2.cxx,v 1.10 2006/10/10 17:43:52 andreas.hoecker Exp $
+// @(#)root/tmva $Id: TSpline2.cxx,v 1.11 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/Timer.cxx b/tmva/src/Timer.cxx
index 7c25018fd2a..288ca66750e 100644
--- a/tmva/src/Timer.cxx
+++ b/tmva/src/Timer.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Timer.cxx,v 1.13 2006/10/15 22:34:22 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Timer.cxx,v 1.14 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
diff --git a/tmva/src/Tools.cxx b/tmva/src/Tools.cxx
index 8233e5541d0..ab70ba6dd85 100644
--- a/tmva/src/Tools.cxx
+++ b/tmva/src/Tools.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Tools.cxx,v 1.52 2006/11/12 15:45:42 stelzer Exp $
+// @(#)root/tmva $Id: Tools.cxx,v 1.55 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -56,6 +56,8 @@ namespace TMVA {
TMVA::MsgLogger& TMVA::Tools::Logger()
{
+ // static access to a common MsgLogger
+
return Tools_Logger ? *Tools_Logger : *(Tools_Logger = new MsgLogger( Tools_NAME_ ));
}
@@ -452,6 +454,8 @@ int TMVA::Tools::GetIndexMinElement(vector<Double_t> &v)
// check if regular expression
Bool_t TMVA::Tools::ContainsRegularExpression( const TString& s )
{
+ // helper function to search for "!%^&()'<>?= " in a string
+
Bool_t regular = kFALSE;
for (Int_t i = 0; i < TMVA::Tools::__regexp__.Length(); i++)
if (s.Contains( TMVA::Tools::__regexp__[i] )) { regular = kTRUE; break; }
@@ -462,6 +466,9 @@ Bool_t TMVA::Tools::ContainsRegularExpression( const TString& s )
// replace regular expressions
TString TMVA::Tools::ReplaceRegularExpressions( const TString& s, TString r )
{
+ // helper function to remove all occurences "!%^&()'<>?= " from a string
+ // and replace all ::,*,/,+,- with _M_,_T_,_D_,_P_,_M_ respectively
+
TString snew = s;
for (Int_t i = 0; i < TMVA::Tools::__regexp__.Length(); i++)
snew.ReplaceAll( TMVA::Tools::__regexp__[i], r );
@@ -487,15 +494,15 @@ void TMVA::Tools::FormattedOutput( const TMatrixD& M, const std::vector<TString>
// get length of each variable, and maximum length
UInt_t minL = 7;
UInt_t maxL = minL;
- std::vector<UInt_t> L;
+ std::vector<UInt_t> vLengths;
for (UInt_t ivar=0; ivar<nvar; ivar++) {
- L.push_back(TMath::Max( (UInt_t)V[ivar].Length(), minL ));
- maxL = TMath::Max( L.back(), maxL );
+ vLengths.push_back(TMath::Max( (UInt_t)V[ivar].Length(), minL ));
+ maxL = TMath::Max( vLengths.back(), maxL );
}
// count column length
UInt_t clen = maxL+1;
- for (UInt_t icol=0; icol<nvar; icol++) clen += L[icol]+1;
+ for (UInt_t icol=0; icol<nvar; icol++) clen += vLengths[icol]+1;
// bar line
for (UInt_t i=0; i<clen; i++) logger << "-";
@@ -503,14 +510,14 @@ void TMVA::Tools::FormattedOutput( const TMatrixD& M, const std::vector<TString>
// title bar
logger << setw(maxL+1) << " ";
- for (UInt_t icol=0; icol<nvar; icol++) logger << setw(L[icol]+1) << V[icol];
+ for (UInt_t icol=0; icol<nvar; icol++) logger << setw(vLengths[icol]+1) << V[icol];
logger << Endl;
// the numbers
for (UInt_t irow=0; irow<nvar; irow++) {
logger << setw(maxL) << V[irow] << ":";
for (UInt_t icol=0; icol<nvar; icol++) {
- logger << setw(L[icol]+1) << Form( "%+1.3f", M(irow,icol) );
+ logger << setw(vLengths[icol]+1) << Form( "%+1.3f", M(irow,icol) );
}
logger << Endl;
}
diff --git a/tmva/src/Types.cxx b/tmva/src/Types.cxx
index 6c792b391b3..b4f01f0c9d5 100644
--- a/tmva/src/Types.cxx
+++ b/tmva/src/Types.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Types.cxx,v 1.8 2006/10/26 19:55:40 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Types.cxx,v 1.11 2006/11/17 00:21:35 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -33,18 +33,21 @@ TMVA::Types* TMVA::Types::fgTypesPtr = 0;
TMVA::Types::Types()
: fLogger( "Types" )
{
- fStr2type["Variable"] = Types::Variable;
- fStr2type["Cuts"] = Types::Cuts;
- fStr2type["Likelihood"] = Types::Likelihood;
- fStr2type["PDERS"] = Types::PDERS;
- fStr2type["HMatrix"] = Types::HMatrix;
- fStr2type["Fisher"] = Types::Fisher;
- fStr2type["CFMlpANN"] = Types::CFMlpANN;
- fStr2type["TMlpANN"] = Types::TMlpANN;
- fStr2type["BDT"] = Types::BDT;
- fStr2type["RuleFit"] = Types::RuleFit;
- fStr2type["SVM"] = Types::SVM;
- fStr2type["MLP"] = Types::MLP;
- fStr2type["BayesClassifier"] = Types::BayesClassifier;
- fStr2type["Committee"] = Types::Committee;
+ // constructor
+ // fill map that links method names and enums
+
+ fStr2type["Variable"] = Types::kVariable;
+ fStr2type["Cuts"] = Types::kCuts;
+ fStr2type["Likelihood"] = Types::kLikelihood;
+ fStr2type["PDERS"] = Types::kPDERS;
+ fStr2type["HMatrix"] = Types::kHMatrix;
+ fStr2type["Fisher"] = Types::kFisher;
+ fStr2type["CFMlpANN"] = Types::kCFMlpANN;
+ fStr2type["TMlpANN"] = Types::kTMlpANN;
+ fStr2type["BDT"] = Types::kBDT;
+ fStr2type["RuleFit"] = Types::kRuleFit;
+ fStr2type["SVM"] = Types::kSVM;
+ fStr2type["MLP"] = Types::kMLP;
+ fStr2type["BayesClassifier"] = Types::kBayesClassifier;
+ fStr2type["Committee"] = Types::kCommittee;
}
diff --git a/tmva/src/VariableInfo.cxx b/tmva/src/VariableInfo.cxx
index 0c227e2cfe4..203696041b4 100644
--- a/tmva/src/VariableInfo.cxx
+++ b/tmva/src/VariableInfo.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: VariableInfo.cxx,v 1.12 2006/10/18 01:24:53 armske Exp $
+// @(#)root/tmva $Id: VariableInfo.cxx,v 1.14 2006/11/17 14:59:24 stelzer Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss
/**********************************************************************************
@@ -13,12 +13,12 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Joerg Stelzer <Joerg.Stelzer@cern.ch> - CERN, Switzerland *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -79,7 +79,7 @@ TMVA::VariableInfo& TMVA::VariableInfo::operator=(const TMVA::VariableInfo& rhs)
}
//_______________________________________________________________________
-void TMVA::VariableInfo::WriteToStream(std::ostream& o, Types::PreprocessingMethod corr) const
+void TMVA::VariableInfo::WriteToStream(std::ostream& o, Types::EPreprocessingMethod corr) const
{
// write VariableInfo to stream
UInt_t nc = TMath::Max( 30, TMath::Max( GetExpression().Length()+1, GetInternalVarName().Length()+1 ) );
@@ -91,7 +91,7 @@ void TMVA::VariableInfo::WriteToStream(std::ostream& o, Types::PreprocessingMeth
}
//_______________________________________________________________________
-void TMVA::VariableInfo::ReadFromStream(std::istream& istr, Types::PreprocessingMethod corr)
+void TMVA::VariableInfo::ReadFromStream(std::istream& istr, Types::EPreprocessingMethod corr)
{
// write VariableInfo to stream
TString exp, varname, vartype, minmax, minstr, maxstr;
diff --git a/tmva/src/Volume.cxx b/tmva/src/Volume.cxx
index 1878c17e93b..7380aca5584 100644
--- a/tmva/src/Volume.cxx
+++ b/tmva/src/Volume.cxx
@@ -1,4 +1,4 @@
-// @(#)root/tmva $Id: Volume.cxx,v 1.9 2006/10/15 22:34:22 andreas.hoecker Exp $
+// @(#)root/tmva $Id: Volume.cxx,v 1.10 2006/11/16 22:51:59 helgevoss Exp $
// Author: Andreas Hoecker, Joerg Stelzer, Helge Voss, Kai Voss
/**********************************************************************************
@@ -13,13 +13,13 @@
* Authors (alphabetical): *
* Andreas Hoecker <Andreas.Hocker@cern.ch> - CERN, Switzerland *
* Xavier Prudent <prudent@lapp.in2p3.fr> - LAPP, France *
- * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Germany *
+ * Helge Voss <Helge.Voss@cern.ch> - MPI-K Heidelberg, Germany *
* Kai Voss <Kai.Voss@cern.ch> - U. of Victoria, Canada *
* *
* Copyright (c) 2005: *
* CERN, Switzerland, *
* U. of Victoria, Canada, *
- * MPI-KP Heidelberg, Germany, *
+ * MPI-K Heidelberg, Germany , *
* LAPP, Annecy, France *
* *
* Redistribution and use in source and binary forms, with or without *
@@ -27,7 +27,7 @@
* (http://tmva.sourceforge.net/LICENSE) *
* *
* File and Version Information: *
- * $Id: Volume.cxx,v 1.9 2006/10/15 22:34:22 andreas.hoecker Exp $
+ * $Id: Volume.cxx,v 1.10 2006/11/16 22:51:59 helgevoss Exp $
**********************************************************************************/
#include "TMVA/Volume.h"
diff --git a/tmva/test/BDT.C b/tmva/test/BDT.C
new file mode 100644
index 00000000000..d057f73333b
--- /dev/null
+++ b/tmva/test/BDT.C
@@ -0,0 +1,189 @@
+#include "tmvaglob.C"
+
+// this macro displays a decision tree read in from the weight file
+
+
+// input: - No. of tree
+// - the weight file from which the tree is read
+void BDT(Int_t itree=1, TString fname= "weights/MVAnalysis_BDT.weights.txt") {
+ draw_tree(itree,fname);
+}
+
+
+//_______________________________________________________________________
+void draw_tree(Int_t itree, TString fname= "weights/MVAnalysis_BDT.weights.txt"){
+ // returns poiter to 2D histogram with tree drawing in it.
+
+ TString *vars;
+ TMVA::DecisionTree *d = read_tree(vars,itree,fname);
+
+ Int_t depth = d->GetDepth();
+ Double_t xmax= 2*depth + 0.5;
+ Double_t xmin= -xmax;
+ Double_t ystep = 1./(depth+1);
+
+ char buffer[100];
+ sprintf (buffer, "Decision Tree No.: %d",itree);
+ TCanvas *c1=new TCanvas("c1",buffer,0,0,1000,600);
+ c1->Draw();
+
+ draw_node( (TMVA::DecisionTreeNode*)d->GetRoot(), 0.5, 1.-0.5*ystep, 0.25, ystep ,vars);
+
+ // make the legend
+ Double_t yup=0.99;
+ Double_t ydown=yup-ystep/2.5;
+ Double_t dy= ystep/2.5 * 0.2;
+
+
+ TPaveText *whichTree = new TPaveText(0.85,ydown,0.98,yup, "NDC");
+ whichTree->SetBorderSize(1);
+ whichTree->SetFillStyle(1);
+ whichTree->SetFillColor(5);
+ whichTree->AddText(buffer);
+ whichTree->Draw();
+
+ TPaveText *intermediate = new TPaveText(0.02,ydown,0.15,yup, "NDC");
+ intermediate->SetBorderSize(1);
+ intermediate->SetFillStyle(1);
+ intermediate->SetFillColor(3);
+ intermediate->AddText("Intermediate Nodes");
+ intermediate->Draw();
+
+
+
+ ydown = ydown - ystep/2.5 -dy;
+ yup = yup - ystep/2.5 -dy;
+ TPaveText *signalleaf = new TPaveText(0.02,ydown ,0.15,yup, "NDC");
+ signalleaf->SetBorderSize(1);
+ signalleaf->SetFillStyle(1);
+ signalleaf->SetFillColor(4);
+ signalleaf->SetTextColor(10);
+ signalleaf->AddText("Signal Leaf Nodes");
+ signalleaf->Draw();
+
+ ydown = ydown - ystep/2.5 -dy;
+ yup = yup - ystep/2.5 -dy;
+ TPaveText *backgroundleaf = new TPaveText(0.02,ydown,0.15,yup, "NDC");
+ backgroundleaf->SetBorderSize(1);
+ backgroundleaf->SetFillStyle(1);
+ backgroundleaf->SetFillColor(2);
+ backgroundleaf->AddText("Background Leaf Nodes");
+ backgroundleaf->Draw();
+
+
+
+}
+
+
+
+//_______________________________________________________________________
+void draw_node( TMVA::DecisionTreeNode *n,
+ Double_t x, Double_t y,
+ Double_t xscale, Double_t yscale, TString * vars) {
+ // recursively puts an entries in the histogram for the node and its daughters
+ //
+
+ if (n->GetLeft() != NULL){
+ TLine *a1 = new TLine(x-xscale/2,y,x-xscale,y-yscale/2);
+ a1->SetLineWidth(2);
+ a1->Draw();
+ draw_node((TMVA::DecisionTreeNode*) n->GetLeft(), x-xscale, y-yscale, xscale/2, yscale, vars);
+ }
+ if (n->GetRight() != NULL){
+ TLine *a1 = new TLine(x+xscale/2,y,x+xscale,y-yscale/2);
+ a1->SetLineWidth(2);
+ a1->Draw();
+ draw_node((TMVA::DecisionTreeNode*) n->GetRight(), x+xscale, y-yscale, xscale/2, yscale, vars );
+ }
+
+
+ TPaveText *t = new TPaveText(x-xscale/2,y-yscale/2,x+xscale/2,y+yscale/2, "NDC");
+
+ t->SetBorderSize(1);
+
+ t->SetFillStyle(1);
+ if (n->GetNodeType() == 1) { t->SetFillColor(4); t->SetTextColor(10); }
+ else if (n->GetNodeType() == -1) t->SetFillColor(2);
+ else if (n->GetNodeType() == 0) t->SetFillColor(3);
+
+ char buffer[25];
+ sprintf(buffer,"N=%d",n->GetNEvents());
+ t->AddText(buffer);
+ sprintf(buffer,"S/(S+B)=%4.3f",n->GetSoverSB());
+ t->AddText(buffer);
+
+ if (n->GetNodeType() == 0){
+ t->AddText(TString(vars[n->GetSelector()]+">"+=::Form("%5.3g",n->GetCutValue())));
+ }
+
+// sprintf(buffer,"seq=%d",n->GetSequence());
+// t->AddText(buffer);
+// sprintf(buffer,"depth=%d",n->GetDepth());
+// t->AddText(buffer);
+ sprintf(buffer,"type=%d",n->GetNodeType());
+ t->AddText(buffer);
+
+
+ if (n->GetNodeType() == 1) t->SetFillColor(4);
+ else if (n->GetNodeType() == -1) t->SetFillColor(2);
+ else if (n->GetNodeType() == 0) t->SetFillColor(3);
+
+
+
+ t->Draw();
+
+ return;
+}
+
+
+TMVA::DecisionTree* read_tree(TString * &vars, Int_t itree=1, TString fname= "weights/MVAnalysis_BDT.weights.txt")
+{
+ cout << "reading Tree " << itree << " from weight file: " << fname << endl;
+ ifstream fin( fname );
+ if (!fin.good( )) { // file not found --> Error
+ cout << "Error opening " << fname << endl;
+ exit(1);
+ }
+
+ Int_t idummy;
+ Float_t fdummy;
+ TString dummy = "";
+
+ // file header with name
+ while (!dummy.Contains("#VAR")) fin >> dummy;
+ fin >> dummy >> dummy >> dummy; // the rest of header line
+
+ // number of variables
+ Int_t nVars;
+ fin >> dummy >> nVars;
+ // at this point, we should have idummy == nVars
+ // cout << "rread nVars = " << nVars <<endl;
+
+
+ // variable mins and maxes
+ vars = new TString[nVars];
+ for (Int_t i = 0; i < nVars; i++) fin >> vars[i] >> dummy >> dummy >> dummy;
+
+ char buffer[20];
+ char line[256];
+ sprintf(buffer,"T %d",itree);
+
+ while (!dummy.Contains(buffer)) {
+ fin.getline(line,256);
+ dummy = TString(line);
+ }
+
+ TMVA::DecisionTreeNode *n = new TMVA::DecisionTreeNode();
+ char pos="s";
+ UInt_t depth =0;
+ n->ReadRec(fin,pos,depth);
+ TMVA::DecisionTree *d = new TMVA::DecisionTree(n);
+
+ // d->Print(cout);
+
+
+ fin.close();
+
+ return d;
+}
+
diff --git a/tmva/test/TMVAGui.C b/tmva/test/TMVAGui.C
new file mode 100644
index 00000000000..5c02ef0bdb2
--- /dev/null
+++ b/tmva/test/TMVAGui.C
@@ -0,0 +1,109 @@
+#include <iostream>
+
+#include "TControlBar.h"
+
+void TMVAGui( const char* fName = "TMVA.root" )
+{
+ // Use this script in order to run the various individual macros
+ // that plot the output of TMVA (e.g. running TMVAnalysis.C),
+ // stored in the file "TMVA.root"
+ // for further documentation, look in the individual macros
+
+
+ cout << "--- Open TMVAGui for input file: " << fName << endl;
+
+ // gROOT->Reset();
+ // gStyle->SetScreenFactor(2); // if you have a large screen, select 1,2 or 1.4
+
+ // create the control bar
+ TControlBar * cbar = new TControlBar( "vertical", "Plotting Scripts", 0, 0 );
+
+ const char* buttonType = "button";
+
+ // configure buttons
+ cbar->AddButton( "Input Variables",
+ Form(".x variables.C(\"%s\",0)",fName),
+ "Plots all input variables (macro variables.C)",
+ buttonType );
+
+ cbar->AddButton( "Decorrelated Variables",
+ Form(".x variables.C(\"%s\",1)",fName),
+ "Plots all decorrelated input variables (macro variables.C)",
+ buttonType );
+
+ cbar->AddButton( "PCA-transformed Variables",
+ Form(".x variables.C(\"%s\",2)",fName),
+ "Plots all PCA-transformed input variables (macro variables.C)",
+ buttonType );
+
+ cbar->AddButton( "Variable Correlations (scatter profiles)",
+ Form(".x correlationscatters.C\(\"%s\",0)",fName),
+ "Plots signal and background correlation profiles between all input variables (macro correlationscatters.C)",
+ buttonType );
+
+ cbar->AddButton( " Decorrelated-Variable Correlations (scatter profiles) ",
+ Form(".x correlationscatters.C\(\"%s\",1)",fName),
+ "Plots signal and background correlation profiles between all decorrelated input variables (macro correlationscatters.C(1))",
+ buttonType );
+
+ cbar->AddButton( " PCA-transformed Variable Correlations (scatter profiles) ",
+ Form(".x correlationscatters.C\(\"%s\",2)",fName),
+ "Plots signal and background correlation profiles between all PCA-transformed input variables (macro correlationscatters.C(2))",
+ buttonType );
+
+ cbar->AddButton( "Variable Correlations (summary)",
+ Form(".x correlations.C(\"%s\")",fName),
+ "Plots signal and background correlation summaries for all input variables (macro correlations.C)",
+ buttonType );
+
+ cbar->AddButton( "Output MVA Variables",
+ Form(".x mvas.C(\"%s\")",fName),
+ "Plots the output variable of each method (macro mvas.C)",
+ buttonType );
+
+ cbar->AddButton( "Mu-transforms (summary)",
+ Form(".x mutransform.C(\"%s\")",fName),
+ "Plots the mu-transformed signal and background MVAs of each method (macro mutransform.C)",
+ buttonType );
+
+ cbar->AddButton( "Background Rejection vs Signal Efficiencies",
+ Form(".x efficiencies.C(\"%s\")",fName),
+ "Plots background rejection vs signal efficiencies (macro efficiencies.C)",
+ buttonType );
+
+ cbar->AddButton( "Likelihood Reference Distributiuons (if exist)",
+ Form(".x likelihoodrefs.C(\"%s\")",fName),
+ "Plots to verify the likelihood reference distributions (macro likelihoodrefs.C)",
+ buttonType );
+
+ cbar->AddButton( "Network Architecture (if exists)",
+ Form(".x network.C(\"%s\")",fName),
+ "Plots the MLP weights (macro network.C)",
+ buttonType );
+
+ cbar->AddButton( "Network Convergence Test (if exists)",
+ Form(".x annconvergencetest.C(\"%s\")",fName),
+ "Plots error estimator versus training epoch for training and test samples (macro annconvergencetest.C)",
+ buttonType );
+
+ cbar->AddButton( "Decision Tree (#1)",
+ Form(".x BDT.C",fName),
+ "Plots the Decision Tree (#1); to plot other trees (i) call macro BDT.C(i) from command line",
+ buttonType );
+
+
+ cbar->AddButton( "Quit", ".q", "Quit", buttonType );
+
+ // set the style
+ cbar->SetTextColor("black");
+
+ // there seems to be a bug in ROOT: font jumps back to default after pressing on >2 different buttons
+ // cbar->SetFont("-adobe-helvetica-bold-r-*-*-12-*-*-*-*-*-iso8859-1");
+
+ // draw
+ cbar->Show();
+
+
+
+ gROOT->SaveContext();
+}
diff --git a/tmva/test/TMVAlogon.C b/tmva/test/TMVAlogon.C
new file mode 100644
index 00000000000..44f2dacc68c
--- /dev/null
+++ b/tmva/test/TMVAlogon.C
@@ -0,0 +1,57 @@
+
+{
+
+ gSystem->Load("libMLP.so");
+
+ // load TMVA shared library created in local release
+ TString libTMVA( "../lib/libTMVA.1.so" );
+ gSystem->Load( libTMVA );
+
+ // welcome the user
+ TMVA::Tools::TMVAWelcomeMessage();
+ cout << "TMVAlogon: loaded TMVA library: \"" << libTMVA << "\"" << endl;
+
+ // some basic style settings
+ TStyle *TMVAStyle = gROOT->GetStyle("Plain"); // our style is based on Plain
+ // new TStyle("TMVA","TMVA plots style");
+ // the pretty color palette of old
+ TMVAStyle->SetPalette(1,0);
+
+ // use plain black on white colors
+ TMVAStyle->SetFrameBorderMode(0);
+ TMVAStyle->SetCanvasBorderMode(0);
+ TMVAStyle->SetPadBorderMode(0);
+ TMVAStyle->SetPadColor(0);
+ TMVAStyle->SetCanvasColor(0);
+ TMVAStyle->SetTitleFillColor(0);
+ TMVAStyle->SetFillStyle(0);
+ TMVAStyle->SetLegendBorderSize(0);
+
+ // set the paper & margin sizes
+ TMVAStyle->SetPaperSize(20,26);
+ TMVAStyle->SetPadTopMargin(0.10);
+ TMVAStyle->SetPadRightMargin(0.05);
+ TMVAStyle->SetPadBottomMargin(0.11);
+ TMVAStyle->SetPadLeftMargin(0.12);
+
+ // use bold lines and markers
+ TMVAStyle->SetMarkerStyle(21);
+ TMVAStyle->SetMarkerSize(0.3);
+ TMVAStyle->SetHistLineWidth(1.85);
+ TMVAStyle->SetLineStyleString(2,"[12 12]"); // postscript dashes
+
+ // do not display any of the standard histogram decorations
+ TMVAStyle->SetOptTitle(1);
+ TMVAStyle->SetTitleH(0.052);
+
+ TMVAStyle->SetOptStat(0);
+ TMVAStyle->SetOptFit(0);
+
+ // put tick marks on top and RHS of plots
+ TMVAStyle->SetPadTickX(1);
+ TMVAStyle->SetPadTickY(1);
+
+ gROOT->SetStyle("Plain");
+ cout << "TMVAlogon: use " << gStyle->GetName() << " style with \"Pretty Color Palette\"" << endl;
+ cout << endl;
+}
diff --git a/tmva/test/TMVAnalysis.C b/tmva/test/TMVAnalysis.C
index 5d2007e0837..6b449a95016 100644
--- a/tmva/test/TMVAnalysis.C
+++ b/tmva/test/TMVAnalysis.C
@@ -1,71 +1,60 @@
+// @(#)root/tmva $Id: TMVAnalysis.C,v 1.62 2006/11/17 00:21:35 stelzer Exp $
/**********************************************************************************
* Project : TMVA - a Root-integrated toolkit for multivariate data analysis *
* Package : TMVA *
* Root Macro: TMVAnalysis *
* *
- * This exectutable provides an example on training and testing of several *
- * Multivariate Analyser (MVA) methods *
+ * This macro gives an example on training and testing of several *
+ * Multivariate Analyser (MVA) methods. *
* *
- * As input file we use a standard Root example of a Higgs analysis, which *
- * is part of the Root tutorial (and hence present in any Root release) *
+ * As input file we use a toy MC sample (you find it in TMVA/examples/data). *
* *
- * The methods to be used can be switched on and off by means of the boolians *
- * below *
+ * The methods to be used can be switched on and off by means of booleans. *
* *
* The output file "TMVA.root" can be analysed with the use of dedicated *
- * macros (simply say: root -l <macro.C>) : *
- * *
- * - variables.C ==> show us the MVA input variables for signal and backgr *
- * - correlations.C ==> show us the correlations between the MVA input vars *
- * - mvas.C ==> show the trained MVAs for the test events *
- * - efficiencies.C ==> show us the background rejections versus signal effs *
- * for all MVAs used *
- * *
- * TMVA allows to train and test multiple MVAs for different phase space *
- * regions. This is however not realised in this simple example. *
- * *
+ * macros (simply say: root -l <macro.C>), which can be conveniently *
+ * invoked through a GUI that will appear at the end of the run of this macro. *
**********************************************************************************/
-void TMVAnalysis() {
-
- // initialisation
- gSystem->Load("libMLP"); // ROOT's Multilayer Perceptron library is needed
- gSystem->Load("libTMVA"); // and of course the TMVA library
- gROOT->ProcessLine(".L loader.C+");
-
- // ---------------------------------------------------------------
- // choose MVA methods to be trained + tested
- Bool_t Use_Cuts = 1;
- Bool_t Use_Likelihood = 1;
- Bool_t Use_LikelihoodD = 1;
- Bool_t Use_PDERS = 1;
- Bool_t Use_HMatrix = 1;
- Bool_t Use_Fisher = 1;
- Bool_t Use_CFMlpANN = 1;
- Bool_t Use_TMlpANN = 1;
- Bool_t Use_BDT_GiniIndex = 1; // default BDT method
- Bool_t Use_BDT_CrossEntro = 0;
- Bool_t Use_BDT_SdivStSpB = 0;
- Bool_t Use_BDT_MisClass = 0;
- Bool_t Use_BDT_Bagging_Gini= 0;
- // ---------------------------------------------------------------
- Bool_t EvaluateVariables = 0; // perform evaluation for each input variable
- // ---------------------------------------------------------------
+#include <iostream>
+
+#include "TCut.h"
+#include "TFile.h"
+#include "TSystem.h"
+#include "TTree.h"
+
+#include "TMVAGui.C"
+
+// ---------------------------------------------------------------
+// choose MVA methods to be trained + tested
+Bool_t Use_Cuts = 1;
+Bool_t Use_CutsD = 0;
+Bool_t Use_Likelihood = 1;
+Bool_t Use_LikelihoodD = 1; // the "D" extension indicates decorrelated input variables (see option strings)
+Bool_t Use_PDERS = 1;
+Bool_t Use_PDERSD = 0;
+Bool_t Use_HMatrix = 1;
+Bool_t Use_Fisher = 1;
+Bool_t Use_MLP = 1; // this is the recommended ANN
+Bool_t Use_CFMlpANN = 0;
+Bool_t Use_TMlpANN = 0;
+Bool_t Use_BDT = 1;
+Bool_t Use_BDTD = 0;
+Bool_t Use_RuleFit = 0;
+
+// read input data file with ascii format (otherwise ROOT) ?
+Bool_t ReadDataFromAsciiIFormat = kFALSE;
+
+void TMVAnalysis()
+{
+ // explicit loading of the shared libTMVA is done in TMVAlogon.C, defined in .rootrc
+ // if you use your private .rootrc, or run from a different directory, please copy the
+ // corresponding lines from .rootrc
- cout << "Start Test TMVAnalysis" << endl
- << "======================" << endl
- << endl;
- cout << "Testing all methods takes about 10-20 minutes. By excluding" << endl
- << "some of the computing expensive analysis methods the demonstration" << endl
- << "will finish much faster." << endl
- << " 1) All methods (15 min)" << endl
- << " 2) Fast methods only (1 min)" << endl
- << "Your choice (1:default): " << flush;
- int selection=1;
- char selc='1';
- cin.get(selc);
- if(selc=='2')
- Use_Cuts = Use_Likelihood = Use_PDERS = Use_HMatrix = Use_CFMlpANN = Use_TMlpANN = Use_BDT_GiniIndex = 0;
+ std::cout << "Start Test TMVAnalysis" << std::endl
+ << "======================" << std::endl
+ << std::endl;
+ std::cout << "Testing all standard methods may take about 4 minutes..." << std::endl;
// Create a new root output file.
TFile* outputFile = TFile::Open( "TMVA.root", "RECREATE" );
@@ -73,182 +62,168 @@ void TMVAnalysis() {
// Create the factory object. Later you can choose the methods whose performance
// you'd like to investigate. The factory will then run the performance analysis
// for you.
- TMVA::Factory *factory = new TMVA::Factory( "MVAnalysis", outputFile, "" ) ;
-
- // Define the signal and background event samples.
- TFile *input(0);
- const char *fname = "tmva_example.root";
- TFile *input = 0;
- if (!gSystem->AccessPathName(fname)) {
- input = TFile::Open(fname);
- } else {
- printf("accessing %s file from http://root.cern.ch/files\n",fname);
- input = TFile::Open(Form("http://root.cern.ch/files/%s",fname));
+ TMVA::Factory *factory = new TMVA::Factory( "MVAnalysis", outputFile, "" );
+
+ if (ReadDataFromAsciiIFormat) {
+ // load the signal and background event samples from ascii files
+ // format in file must be:
+ // var1/F:var2/F:var3/F:var4/F
+ // 0.04551 0.59923 0.32400 -0.19170
+ // ...
+
+ TString datFileS = "data/toy_sig_lincorr.dat";
+ TString datFileB = "data/toy_bkg_lincorr.dat";
+ if (!factory->SetInputTrees( datFileS, datFileB )) exit(1);
}
- if (!input) return;
-
- TTree *signal = (TTree*)input->Get("TreeS");
- TTree *background = (TTree*)input->Get("TreeB");
- if( ! factory->SetInputTrees( signal, background )) return;
-
-
- // Define the input variables. These are used in the TMVA.
- vector<TString>* inputVars = new vector<TString>;
- inputVars->push_back("var1");
- inputVars->push_back("var2");
- inputVars->push_back("var3");
- inputVars->push_back("var4");
- factory->SetInputVariables( inputVars );
-
-
+ else {
+ // load the signal and background event samples from ROOT trees
+ TFile *input(0);
+ if (!gSystem->AccessPathName("../examples/data/toy_sigbkg.root")) {
+ // ../examples/data/toy_sigbkg.root is
+ // available with the sourceforge installation
+ cout << "--- accessing ../examples/data/toy_sigbkg.root" << endl;
+ input = TFile::Open("../examples/data/toy_sigbkg.root");
+ } else if(!gSystem->AccessPathName("tmva_example.root")) {
+ // if you downloaded the example from the ROOT site
+ // http://root.cern.ch/files/tmva_example.root
+ // into the local directory
+ cout << "--- accessing ./tmva_example.root" << endl;
+ input = TFile::Open("tmva_example.root");
+ } else {
+ // try accessing the file via the web from the ROOT site
+ cout << "--- accessing tmva_example.root file from http://root.cern.ch/files" << endl;
+ cout << "--- for faster startup you may consider downloading it into you local directory" << endl;
+ input = TFile::Open("http://root.cern.ch/files/tmva_example.root");
+ }
+
+ if (!input) {
+ std::cout << "ERROR: could not open data file" << std::endl;
+ exit(1);
+ }
+
+ TTree *signal = (TTree*)input->Get("TreeS");
+ TTree *background = (TTree*)input->Get("TreeB");
+
+ // global event weights (see below for setting event-wise weights)
+ Double_t signalWeight = 1.0;
+ Double_t backgroundWeight = 1.0;
+
+ // sanity check
+ if (!signal || !background) {
+ std::cout << "ERROR: unknown tree(s)" << std::endl;
+ exit(1);
+ }
+ if (!factory->SetInputTrees( signal, background, signalWeight, backgroundWeight)) exit(1);
+ }
+
+ // Define the input variables that shall be used for the MVA training
+ // note that you may also use variable expressions, such as: "3*var1/var2*abs(var3)"
+ // [all types of expressions that can also be parsed by TTree::Draw( "expression" )]
+ factory->AddVariable("var1", 'F');
+ factory->AddVariable("var2", 'F');
+ factory->AddVariable("var3", 'F');
+ factory->AddVariable("var4", 'F');
+
+ // This would set individual event weights (the variables defined in the
+ // expression need to exist in the original TTree)
+ // factory->SetWeightExpression("weight1*weight2");
+
// Apply additional cuts on the signal and background sample.
// Assumptions on size of training and testing sample:
// a) equal number of signal and background events is used for training
// b) any numbers of signal and background events are used for testing
// c) an explicit syntax can violate a)
// more Documentation with the Factory class
- TCut mycut = "";
- factory->PrepareTrainingAndTestTree( mycut, 2000, 4000 );
+ TCut mycut = ""; // for example: TCut mycut = "abs(var1)<0.5 && abs(var2-0.5)<1";
+
+ factory->PrepareTrainingAndTestTree( mycut, 1000, 40000 );
+ // ---- book MVA methods
+ //
+ // please lookup the various method configuration options in the corresponding cxx files, eg:
+ // src/MethoCuts.cxx, etc.
- // Book the MVA methods you like to investigate.
-
- // MethodCuts:
- // format of option string: "OptMethod:EffMethod:Option_var1:...:Option_varn"
- // "OptMethod" can be:
- // - "GA" : Genetic Algorithm (recommended)
- // - "MC" : Monte-Carlo optimization
- // "EffMethod" can be:
- // - "EffSel": compute efficiency by event counting
- // - "EffPDF": compute efficiency from PDFs
- // === For "GA" method ======
- // "Option_var1++" are (see GA for explanation of parameters):
- // - fGa_nsteps
- // - fGa_preCalc
- // - fGa_SC_steps
- // - fGa_SC_offsteps
- // - fGa_SC_factor
- // === For "MC" method ======
- // "Option_var1" is number of random samples
- // "Option_var2++" can be
- // - "FMax" : ForceMax (the max cut is fixed to maximum of variable i)
- // - "FMin" : ForceMin (the min cut is fixed to minimum of variable i)
- // - "FSmart": ForceSmart (the min or max cut is fixed to min/max, based on mean value)
- // - Adding "All" to "option_vari", eg, "AllFSmart" will use this option for all variables
- // - if "option_vari" is empty (== ""), no assumptions on cut min/max are made
- // ----------------------------------------------------------------------------------
+ // Cut optimisation
if (Use_Cuts)
- factory->BookMethod( "MethodCuts", "V:GA:EffSel:30:3:10:5:0.95" );
- // factory->BookMethod( "MethodCuts", "V:MC:EffSel:10000:AllFSmart" );
-
- // MethodLikelihood options:
- // histogram_interpolation_method:nsmooth:nsmooth:n_aveEvents_per_bin:Decorrelation
+ factory->BookMethod( TMVA::Types::kCuts, "Cuts", "!V:MC:EffSel:MC_NRandCuts=100000:AllFSmart" );
+
+ // alternatively, use the powerful cut optimisation with a Genetic Algorithm
+ // factory->BookMethod( TMVA::Types::Cuts, "CutsGA",
+ // "!V:GA:EffSel:GA_nsteps=40:GA_cycles=30:GA_popSize=100:GA_SC_steps=10:GA_SC_offsteps=5:GA_SC_factor=0.95" );
+
+ if (Use_CutsD)
+ factory->BookMethod( TMVA::Types::kCuts, "CutsD", "!V:MC:EffSel:MC_NRandCuts=200000:AllFSmart:Preprocess=Decorrelate" );
+
+ // Likelihood
if (Use_Likelihood)
- factory->BookMethod( TMVA::Types::Likelihood, "Spline2:3" );
- if (Use_LikelihoodD)
- factory->BookMethod( TMVA::Types::Likelihood, "Spline2:10:25:D");
+ factory->BookMethod( TMVA::Types::kLikelihood, "Likelihood", "!V:!TransformOutput:Spline=2:NSmooth=5" );
+
+ // test the decorrelated likelihood
+ if (Use_LikelihoodD)
+ factory->BookMethod( TMVA::Types::kLikelihood, "LikelihoodD", "!V:!TransformOutput:Spline=2:NSmooth=5:Preprocess=Decorrelate");
- // MethodFisher:
+ // Fisher:
if (Use_Fisher)
- factory->BookMethod( TMVA::Types::Fisher, "Fisher" ); // Fisher method ("Fi" or "Ma")
-
- // Method CF(Clermont-Ferrand)ANN:
+ factory->BookMethod( TMVA::Types::kFisher, "Fisher", "!V:Fisher" );
+
+ // the new TMVA ANN: MLP (recommended ANN)
+ if (Use_MLP)
+ factory->BookMethod( TMVA::Types::kMLP, "MLP", "!V:NCycles=200:HiddenLayers=N+1,N:TestRate=5" );
+
+ // CF(Clermont-Ferrand)ANN
if (Use_CFMlpANN)
- factory->BookMethod( TMVA::Types::CFMlpANN, "5000:N:N" ); // n_cycles:#nodes:#nodes:...
+ factory->BookMethod( TMVA::Types::kCFMlpANN, "CFMlpANN", "!V:H:NCycles=5000:HiddenLayers=N,N" ); // n_cycles:#nodes:#nodes:...
- // Method CF(Root)ANN:
+ // Tmlp(Root)ANN
if (Use_TMlpANN)
- factory->BookMethod( TMVA::Types::TMlpANN, "200:N+1:N" ); // n_cycles:#nodes:#nodes:...
+ factory->BookMethod( TMVA::Types::kTMlpANN, "TMlpANN", "!V:NCycles=200:HiddenLayers=N+1,N" ); // n_cycles:#nodes:#nodes:...
- // MethodHMatrix:
+ // HMatrix
if (Use_HMatrix)
- factory->BookMethod( TMVA::Types::HMatrix ); // H-Matrix (chi2-squared) method
+ factory->BookMethod( TMVA::Types::kHMatrix, "HMatrix", "!V" ); // H-Matrix (chi2-squared) method
// PDE - RS method
- // format and syntax of option string: "VolumeRangeMode:options"
- // where:
- // VolumeRangeMode - all methods defined in private enum "VolumeRangeMode"
- // options - deltaFrac in case of VolumeRangeMode=MinMax/RMS
- // - nEventsMin/Max, maxVIterations, scale for VolumeRangeMode=Adaptive
if (Use_PDERS)
- factory->BookMethod( TMVA::Types::PDERS, "Adaptive:50:100:50:0.99" );
-
- // MethodBDT (Boosted Decision Trees) options:
- // format and syntax of option string: "nTrees:BoostType:SeparationType:
- // nEventsMin:dummy:
- // nCuts:SignalFraction"
- // nTrees: number of trees in the forest to be created
- // BoostType: the boosting type for the trees in the forest (AdaBoost e.t.c..)
- // SeparationType the separation criterion applied in the node splitting
- // nEventsMin: the minimum number of events in a node (leaf criteria, stop splitting)
- // dummy: dummy option to keep backward compatible
- // continue splitting. !!
- // !!! Needs to be set to zero, as it doesn't work properly otherwise
- // ... it's strange though and not yet understood !!!
- // nCuts: the number of steps in the optimisation of the cut for a node
- // SignalFraction: scale parameter of the number of Bkg events
- // applied to the training sample to simulate different initial purity
- // of your data sample.
- //
- // known SeparationTypes are:
- // - MisClassificationError
- // - GiniIndex
- // - CrossEntropy
- // known BoostTypes are:
- // - AdaBoost
- // - Bagging
-
- if (Use_BDT_GiniIndex)
- factory->BookMethod( TMVA::Types::BDT, "200:AdaBoost:GiniIndex:10:0.:20" );
- if (Use_BDT_CrossEntro)
- factory->BookMethod( TMVA::Types::BDT, "200:AdaBoost:CrossEntropy:10:0.:20" );
- if (Use_BDT_SdivStSpB)
- factory->BookMethod( TMVA::Types::BDT, "200:AdaBoost:SdivSqrtSplusB:10:0.:20" );
- if (Use_BDT_MisClass)
- factory->BookMethod( TMVA::Types::BDT, "200:AdaBoost:MisClassificationError:10:0.:20" );
- if (Use_BDT_Bagging_Gini)
- factory->BookMethod( TMVA::Types::BDT, "200:Bagging:GiniIndex:10:0.:20","bagging" );
+ factory->BookMethod( TMVA::Types::kPDERS, "PDERS",
+ "!V:VolumeRangeMode=RMS:KernelEstimator=Teepee:MaxVIterations=50:InitialScale=0.99" ) ;
+
+ if (Use_PDERSD)
+ factory->BookMethod( TMVA::Types::kPDERS, "PDERSD",
+ "!V:VolumeRangeMode=RMS:KernelEstimator=Teepee:MaxVIterations=50:InitialScale=0.99:Preprocess=Decorrelate" ) ;
+ // Boosted Decision Trees
+ if (Use_BDT)
+ factory->BookMethod( TMVA::Types::kBDT, "BDT",
+ "!V:NTrees=400:BoostType=AdaBoost:SeparationType=GiniIndex:nEventsMin=20:nCuts=20:PruneMethod=CostComplexity:PruneStrength=3.5");
+ if (Use_BDTD)
+ factory->BookMethod( TMVA::Types::kBDT, "BDTD",
+ "!V:NTrees=400:BoostType=AdaBoost:SeparationType=GiniIndex:nEventsMin=20:nCuts=20:PruneMethod=CostComplexity:PruneStrength=3.5:Preprocess=Decorrelate");
+
+ // Friedman's RuleFit method
+ if (Use_RuleFit)
+ factory->BookMethod( TMVA::Types::kRuleFit, "RuleFit",
+ "!V:NTrees=20:SampleFraction=-1:nEventsMin=60:nCuts=20:MinImp=0.001:Model=ModLinear:GDTau=0.6:GDStep=0.01:GDNSteps=100000:SeparationType=GiniIndex:RuleMaxDist=0.00001" );
- // Now you can tell the factory to train, test, and evaluate the MVAs.
+ // ---- Now you can tell the factory to train, test, and evaluate the MVAs.
// Train MVAs.
factory->TrainAllMethods();
-
// Test MVAs.
factory->TestAllMethods();
-
- // Evaluate variables.
- if (EvaluateVariables) factory->EvaluateAllVariables();
-
-
// Evaluate MVAs
factory->EvaluateAllMethods();
-
// Save the output.
outputFile->Close();
-
- cout << "==> wrote root file TMVA.root" << endl;
- cout << "==> TMVAnalysis is done!" << endl;
-
+ std::cout << "==> wrote root file TMVA.root" << std::endl;
+ std::cout << "==> TMVAnalysis is done!" << std::endl;
// clean up
delete factory;
- delete inputVars;
-
- gROOT->Reset();
- gStyle->SetScreenFactor(1); //if you have a large screen, select 1,2 or 1.4
- bar = new TControlBar("vertical", "Checks",0,0);
- bar->AddButton("Input Variables", ".x variables.C", "Plots all input variables (macro variables.C)");
- bar->AddButton("Variable Correlations", ".x correlations.C", "Plots correlations between all input variables (macro variables.C)");
- bar->AddButton("Output MVA Variables", ".x mvas.C", "Plots the output variable of each method (macro mvas.C)");
- bar->AddButton("Background Rejection vs Signal Efficiencies", ".x efficiencies.C", "Plots background rejection vs signal efficiencies (macro efficiencies.C)");
- bar->AddButton("Quit", ".q", "Quit");
- bar->Show();
- gROOT->SaveContext();
+ // open the GUI for the root macros
+ TMVAGui();
}
diff --git a/tmva/test/TMVAnalysis.py b/tmva/test/TMVAnalysis.py
new file mode 100644
index 00000000000..68840bcebd0
--- /dev/null
+++ b/tmva/test/TMVAnalysis.py
@@ -0,0 +1,257 @@
+#!/usr/bin/env python
+# @(#)root/tmva $Id: TMVAnalysis.py,v 1.3 2006/11/15 11:00:51 andreas.hoecker Exp $
+# ------------------------------------------------------------------------------ #
+# Project : TMVA - a Root-integrated toolkit for multivariate data analysis #
+# Package : TMVA #
+# Python script: TMVAnalysis.py #
+# #
+# This python script gives an example on training and testing of several #
+# Multivariate Analyser (MVA) methods through PyROOT. Note that PyROOT requires #
+# that you have a python version > 2.2 installed on your computer. #
+# #
+# As input file we use a toy MC sample (you find it in TMVA/examples/data) #
+# #
+# The methods to be used can be switched on and off by means of booleans. #
+# #
+# The output file "TMVA.root" can be analysed with the use of dedicated #
+# macros (simply say: root -l <../macros/macro.C>), which can be conveniently #
+# invoked through a GUI that will appear at the end of the run of this macro. #
+# #
+# for help type "python TMVAnalysis.py --help" #
+# ------------------------------------------------------------------------------ #
+
+# --------------------------------------------
+# standard python import
+import sys # exit
+import time # time accounting
+import getopt # command line parser
+
+# --------------------------------------------
+
+# default settings for command line arguments
+DEFAULT_OUTFNAME = "TMVA.root"
+DEFAULT_INFNAME = "../examples/data/toy_sigbkg.root"
+DEFAULT_TREESIG = "TreeS"
+DEFAULT_TREEBKG = "TreeB"
+DEFAULT_METHODS = "Cuts CutsD Likelihood LikelihoodD PDERS HMatrix Fisher MLP BDT"
+
+# print help
+def usage():
+ print " "
+ print "Usage: python %s [options]" % sys.argv[0]
+ print " -m | --methods : gives methods to be run (default: all methods)"
+ print " -i | --inputfile : name of input ROOT file (default: '%s')" % DEFAULT_INFNAME
+ print " -o | --outputfile : name of output ROOT file containing results (default: '%s')" % DEFAULT_OUTFNAME
+ print " -t | --inputtrees : input ROOT Trees for signal and background (default: '%s %s')" \
+ % (DEFAULT_TREESIG, DEFAULT_TREEBKG)
+ print " -v | --verbose"
+ print " -? | --usage : print this help message"
+ print " -h | --help : print this help message"
+ print " "
+
+# main routine
+def main():
+
+ try:
+ # retrive command line options
+ shortopts = "m:i:t:o:vh?"
+ longopts = ["methods=", "inputfile=", "inputtrees=", "outputfile=", "verbose", "help", "usage"]
+ opts, args = getopt.getopt( sys.argv[1:], shortopts, longopts )
+
+ except getopt.GetoptError:
+ # print help information and exit:
+ print "ERROR: unknown options in argument %s" % sys.argv[1:]
+ usage()
+ sys.exit(1)
+
+ infname = DEFAULT_INFNAME
+ treeNameSig = DEFAULT_TREESIG
+ treeNameBkg = DEFAULT_TREEBKG
+ outfname = DEFAULT_OUTFNAME
+ methods = DEFAULT_METHODS
+ verbose = False
+ for o, a in opts:
+ if o in ("-?", "-h", "--help", "--usage"):
+ usage()
+ sys.exit(0)
+ elif o in ("-m", "--methods"):
+ methods = a
+ elif o in ("-i", "--inputfile"):
+ infname = a
+ elif o in ("-o", "--outputfile"):
+ outfname = a
+ elif o in ("-t", "--inputtrees"):
+ a.strip()
+ trees = a.rsplit( ' ' )
+ trees.sort()
+ trees.reverse()
+ if len(trees)-trees.count('') != 2:
+ print "ERROR: need to give two trees (each one for signal and background)"
+ print trees
+ sys.exit(1)
+ treeNameSig = trees[0]
+ treeNameBkg = trees[1]
+ elif o in ("-v", "--verbose"):
+ verbose = True
+
+ # print methods
+ mlist = methods.split(' ')
+ print "=== TMVAnalysis: use methods..."
+ for m in mlist:
+ if m != '':
+ print "=== ... <%s>" % m
+
+ # import ROOT classes
+ from ROOT import gSystem, gROOT, gApplication, TFile, TTree, TCut
+
+ # load TMVA library and GUI
+ gSystem.Load( 'libTMVA.1.so' )
+ gROOT.LoadMacro( '../macros/TMVAGui.C' )
+
+ # import TMVA classes from ROOT
+ from ROOT import TMVA
+
+ # output file
+ outputFile = TFile( outfname, 'RECREATE' )
+
+ # create einstance of factory
+ factory = TMVA.Factory( "MVAnalysis", outputFile, "" )
+
+ # set verbosity
+ factory.SetVerbose( verbose )
+
+ # read input data
+ if not gSystem.AccessPathName( infname ):
+ input = TFile( infname )
+ else:
+ print "ERROR: could not access data file %s\n" % infname
+
+ signal = input.Get( treeNameSig )
+ background = input.Get( treeNameBkg )
+
+ # global event weights (see below for setting event-wise weights)
+ signalWeight = 1.0
+ backgroundWeight = 1.0
+
+ if not factory.SetInputTrees( signal, background, signalWeight, backgroundWeight ):
+ print "ERROR: could not set input trees\n"
+ sys.exit(1)
+
+ # Define the input variables that shall be used for the MVA training
+ # note that you may also use variable expressions, such as: "3*var1/var2*abs(var3)"
+ # [all types of expressions that can also be parsed by TTree::Draw( "expression" )]
+ factory.AddVariable("var1", 'F')
+ factory.AddVariable("var2", 'F')
+ factory.AddVariable("var3", 'F')
+ factory.AddVariable("var4", 'F')
+
+ # This would set individual event weights (the variables defined in the
+ # expression need to exist in the original TTree)
+ # factory->SetWeightExpression("weight1*weight2")
+ #
+ # Apply additional cuts on the signal and background sample.
+ # Assumptions on size of training and testing sample:
+ # a) equal number of signal and background events is used for training
+ # b) any numbers of signal and background events are used for testing
+ # c) an explicit syntax can violate a)
+ # more Documentation with the Factory class
+ # example for cut: mycut = TCut( "abs(var1)<0.5 && abs(var2-0.5)<1" )
+ mycut = TCut( "" )
+
+ # here, the relevant variables are copied over in new, slim trees that are
+ # used for TMVA training and testing
+ factory.PrepareTrainingAndTestTree( mycut, 2000, 4000 )
+
+ # Cut optimisation
+ if methods.find( "Cuts" ) != -1:
+ factory.BookMethod( TMVA.Types.Cuts, "Cuts", "!V:MC:EffSel:MC_NRandCuts=100000:AllFSmart" )
+
+ # Cut optimisation with a Genetic Algorithm
+ if methods.find( "Cuts" ) != -1:
+ factory.BookMethod( TMVA.Types.Cuts, "CutsGA",
+ "!V:GA:EffSel:GA_nsteps=40:GA_cycles=30:GA_popSize=100:GA_SC_steps=10:GA_SC_offsteps=5:GA_SC_factor=0.95" )
+
+ # Cut optimisation using decorrelated input variables
+ if methods.find( "CutsD" ) != -1:
+ factory.BookMethod( TMVA.Types.Cuts, "CutsD", "!V:MC:EffSel:MC_NRandCuts=200000:AllFSmart:Preprocess=Decorrelate" )
+
+ # Likelihood
+ if methods.find( "Likelihood" ) != -1:
+ factory.BookMethod( TMVA.Types.Likelihood, "Likelihood", "!V:!TransformOutput:Spline=2:NSmooth=5" )
+
+ # test the decorrelated likelihood
+ if methods.find( "LikelihoodD" ) != -1:
+ factory.BookMethod( TMVA.Types.Likelihood, "LikelihoodD", "!V:!TransformOutput:Spline=2:NSmooth=5:Preprocess=Decorrelate")
+
+ # Fisher:
+ if methods.find( "Fisher" ) != -1:
+ factory.BookMethod( TMVA.Types.Fisher, "Fisher", "!V:Fisher" )
+
+ # the new TMVA ANN: MLP (recommended ANN)
+ if methods.find( "MLP" ) != -1:
+ factory.BookMethod( TMVA.Types.MLP, "MLP", "!V:NCycles=200:HiddenLayers=N+1,N:TestRate=5" )
+
+ # CF(Clermont-Ferrand)ANN
+ if methods.find( "CFMlpANN" ) != -1:
+ factory.BookMethod( TMVA.Types.CFMlpANN, "CFMlpANN", "!V:H:NCycles=5000:HiddenLayers=N,N" ) # n_cycles:#nodes:#nodes:...
+
+ # Tmlp(Root)ANN
+ if methods.find( "TMlpANN" ) != -1:
+ factory.BookMethod( TMVA.Types.TMlpANN, "TMlpANN", "!V:NCycles=200:HiddenLayers=N+1,N" ) # n_cycles:#nodes:#nodes:...
+
+ # HMatrix
+ if methods.find( "HMatrix" ) != -1:
+ factory.BookMethod( TMVA.Types.HMatrix, "HMatrix", "!V" ) # H-Matrix (chi2-squared) method
+
+ # PDE - RS method
+ if methods.find( "PDERS" ) != -1:
+ factory.BookMethod( TMVA.Types.PDERS, "PDERS",
+ "!V:VolumeRangeMode=RMS:KernelEstimator=Teepee:MaxVIterations=50:InitialScale=0.99" )
+ if methods.find( "PDERSD" ) != -1:
+ factory.BookMethod( TMVA.Types.PDERS, "PDERSD",
+ "!V:VolumeRangeMode=RMS:KernelEstimator=Teepee:MaxVIterations=50:InitialScale=0.99:Preprocess=Decorrelate" )
+
+ # Boosted Decision Trees
+ if methods.find( "BDT" ) != -1:
+ factory.BookMethod( TMVA.Types.BDT, "BDT",
+ "!V:NTrees=400:BoostType=AdaBoost:SeparationType=GiniIndex:nEventsMin=20:SignalFraction=0.:nCuts=20:PruneMethod=CostComplexity:PruneStrength=3.5" );
+
+ # Friedman's RuleFit method
+ if methods.find( "RuleFit" ) != -1:
+ factory.BookMethod( TMVA.Types.RuleFit, "RuleFit",
+ "!V:NTrees=20:SampleFraction=-1:nEventsMin=60:nCuts=20:MinImp=0.001:Model=ModLinear:GDTau=0.6:GDStep=0.01:GDNSteps=100000:SeparationType=GiniIndex:RuleMaxDist=0.00001" )
+
+ # Bayesian classifier
+ if methods.find( "BayesClassifier" ) != -1:
+ factory.BookMethod( TMVA.Types.BayesClassifier, "BayesClassifier", "!V:myOptions" )
+
+ # ---- Now you can tell the factory to train, test, and evaluate the MVAs.
+
+ # Train MVAs
+ factory.TrainAllMethods()
+
+ # Test MVAs
+ factory.TestAllMethods()
+
+ # Evaluate MVAs
+ factory.EvaluateAllMethods()
+
+ # Save the output.
+ outputFile.Close()
+
+ # clean up
+ factory.IsA().Destructor( factory )
+
+ print "=== wrote root file %s\n" % outfname
+ print "=== TMVAnalysis is done!\n"
+
+ # open the GUI for the result macros
+ gROOT.ProcessLine( "TMVAGui(\"%s\")" % outfname );
+
+ # keep the ROOT thread running
+ gApplication.Run()
+
+# ----------------------------------------------------------
+
+if __name__ == "__main__":
+ main()
diff --git a/tmva/test/TMVApplication.C b/tmva/test/TMVApplication.C
new file mode 100644
index 00000000000..67c1d8aefc7
--- /dev/null
+++ b/tmva/test/TMVApplication.C
@@ -0,0 +1,176 @@
+/**********************************************************************************
+ * Project : TMVA - a Root-integrated toolkit for multivariate data analysis *
+ * Package : TMVA *
+ * Exectuable: TMVApplication *
+ * *
+ * This exectutable provides a simple example on how to use the trained MVAs *
+ * within a C++ analysis module *
+ * *
+ * ------------------------------------------------------------------------------ *
+ * see also the alternative (slightly faster) way to retrieve the MVA values in *
+ * examples/TMVApplicationAlternative.cxx *
+ * ------------------------------------------------------------------------------ *
+ **********************************************************************************/
+
+// ---------------------------------------------------------------
+// choose MVA methods to be trained + tested
+Bool_t Use_Cuts = 1;
+Bool_t Use_CutsD = 1;
+Bool_t Use_Likelihood = 1;
+Bool_t Use_LikelihoodD = 1; // the "D" extension indicates decorrelated input variables (see option strings)
+Bool_t Use_PDERS = 1;
+Bool_t Use_PDERSD = 1;
+Bool_t Use_HMatrix = 1;
+Bool_t Use_Fisher = 1;
+Bool_t Use_MLP = 1; // this is the recommended ANN
+Bool_t Use_CFMlpANN = 1;
+Bool_t Use_TMlpANN = 1;
+Bool_t Use_BDT = 1;
+Bool_t Use_BDTD = 1;
+Bool_t Use_RuleFit = 1;
+// ---------------------------------------------------------------
+
+void TMVApplication()
+{
+ cout << endl;
+ cout << "==> start TMVApplication" << endl;
+
+ //
+ // create the Reader object
+ //
+ TMVA::Reader *reader = new TMVA::Reader();
+
+ // create a set of variables and declare them to the reader
+ // - the variable names must corresponds in name and type to
+ // those given in the weight file(s) that you use
+ Float_t var1, var2, var3, var4;
+ reader->AddVariable( "var1", &var1 );
+ reader->AddVariable( "var2", &var2 );
+ reader->AddVariable( "var3", &var3 );
+ reader->AddVariable( "var4", &var4 );
+
+ //
+ // book the MVA methods
+ //
+ string dir = "weights/";
+ string prefix = "MVAnalysis";
+
+ if (Use_Cuts) reader->BookMVA( "Cuts method", dir + prefix + "_Cuts.weights.txt" );
+ if (Use_CutsD) reader->BookMVA( "CutsD method", dir + prefix + "_CutsD.weights.txt" );
+ if (Use_Likelihood) reader->BookMVA( "Likelihood method", dir + prefix + "_Likelihood.weights.txt" );
+ if (Use_LikelihoodD) reader->BookMVA( "LikelihoodD method", dir + prefix + "_LikelihoodD.weights.txt" );
+ if (Use_PDERS) reader->BookMVA( "PDERS method", dir + prefix + "_PDERS.weights.txt" );
+ if (Use_PDERSD) reader->BookMVA( "PDERSD method", dir + prefix + "_PDERSD.weights.txt" );
+ if (Use_HMatrix) reader->BookMVA( "HMatrix method", dir + prefix + "_HMatrix.weights.txt" );
+ if (Use_Fisher) reader->BookMVA( "Fisher method", dir + prefix + "_Fisher.weights.txt" );
+ if (Use_MLP) reader->BookMVA( "MLP method", dir + prefix + "_MLP.weights.txt" );
+ if (Use_CFMlpANN) reader->BookMVA( "CFMlpANN method", dir + prefix + "_CFMlpANN.weights.txt" );
+ if (Use_TMlpANN) reader->BookMVA( "TMlpANN method", dir + prefix + "_TMlpANN.weights.txt" );
+ if (Use_BDT) reader->BookMVA( "BDT method", dir + prefix + "_BDT.weights.txt" );
+ if (Use_BDTD) reader->BookMVA( "BDTD method", dir + prefix + "_BDTD.weights.txt" );
+ if (Use_RuleFit) reader->BookMVA( "RuleFit method", dir + prefix + "_RuleFit.weights.txt" );
+
+ //
+ // Prepare input tree (this must be replaced by your data source)
+ // in this example, there is a toy tree with signal and one with background events
+ // we'll later on use only the "signal" events for the test in this example.
+ //
+ TFile *input = new TFile("../examples/data/toy_sigbkg.root");
+ TTree *signal = (TTree*)input->Get("TreeS");
+ TTree *background = (TTree*)input->Get("TreeB");
+
+ //
+ // book output histograms
+ UInt_t nbin = 100;
+ TH1F *histLk, *histLkD, *histPD, *histPDD, *histHm, *histFi, *histNn, *histNnC, *histNnT, *histBdt, *histRf;
+ if (Use_Likelihood) histLk = new TH1F( "MVA_Likelihood", "MVA_Likelihood", nbin, 0, 1 );
+ if (Use_LikelihoodD) histLkD = new TH1F( "MVA_LikelihoodD", "MVA_LikelihoodD", nbin, 0, 1 );
+ if (Use_PDERS) histPD = new TH1F( "MVA_PDERS", "MVA_PDERS", nbin, 0, 1 );
+ if (Use_PDERSD) histPDD = new TH1F( "MVA_PDERSD", "MVA_PDERSD", nbin, 0, 1 );
+ if (Use_HMatrix) histHm = new TH1F( "MVA_HMatrix", "MVA_HMatrix", nbin, -0.95, 1.55 );
+ if (Use_Fisher) histFi = new TH1F( "MVA_Fisher", "MVA_Fisher", nbin, -4, 4 );
+ if (Use_MLP) histNn = new TH1F( "MVA_MLP", "MVA_MLP", nbin, -0.25, 1.5 );
+ if (Use_CFMlpANN) histNnC = new TH1F( "MVA_CFMlpANN", "MVA_CFMlpANN", nbin, 0, 1 );
+ if (Use_TMlpANN) histNnT = new TH1F( "MVA_TMlpANN", "MVA_TMlpANN", nbin, -1, 1 );
+ if (Use_BDT) histBdt = new TH1F( "MVA_BDT", "MVA_BDT", nbin, -0.4, 0.6 );
+ if (Use_BDTD) histBdtD= new TH1F( "MVA_BDTD", "MVA_BDTD", nbin, -0.4, 0.6 );
+ if (Use_RuleFit) histRf = new TH1F( "MVA_RuleFit", "MVA_RuleFit", nbin, -1.3, 1.3 );
+
+ //
+ // prepare the tree
+ // - here the variable names have to corresponds to your tree
+ // - you can use the same variables as above which is slightly faster,
+ // but of course you can use different ones and copy the values inside the event loop
+ //
+ TTree* theTree = signal;
+ theTree->SetBranchAddress( "var1", &var1 );
+ theTree->SetBranchAddress( "var2", &var2 );
+ theTree->SetBranchAddress( "var3", &var3 );
+ theTree->SetBranchAddress( "var4", &var4 );
+
+ // efficiency calculator for cut method
+ int nSelCuts = 0;
+ double effS = 0.7;
+
+ cout << "--- processing: " << theTree->GetEntries() << " events" << endl;
+ for (Long64_t ievt=0; ievt<theTree->GetEntries();ievt++) {
+
+ theTree->GetEntry(ievt);
+
+ if (ievt%1000 == 0)
+ cout << "--- ... processing event: " << ievt << endl;
+
+ //
+ // return the MVA
+ //
+ if (Use_Cuts) {
+ // give the desired signal efficienciy
+ Bool_t passed = reader->EvaluateMVA( "Cuts method", effS );
+ if (passed) nSelCuts++;
+ }
+
+ //
+ // fill histograms
+ //
+ if (Use_Likelihood ) histLk ->Fill( reader->EvaluateMVA( "Likelihood method" ) );
+ if (Use_LikelihoodD) histLkD ->Fill( reader->EvaluateMVA( "LikelihoodD method" ) );
+ if (Use_PDERS ) histPD ->Fill( reader->EvaluateMVA( "PDERS method" ) );
+ if (Use_PDERSD ) histPDD ->Fill( reader->EvaluateMVA( "PDERSD method" ) );
+ if (Use_HMatrix ) histHm ->Fill( reader->EvaluateMVA( "HMatrix method" ) );
+ if (Use_Fisher ) histFi ->Fill( reader->EvaluateMVA( "Fisher method" ) );
+ if (Use_MLP ) histNn ->Fill( reader->EvaluateMVA( "MLP method" ) );
+ if (Use_CFMlpANN ) histNnC ->Fill( reader->EvaluateMVA( "CFMlpANN method" ) );
+ if (Use_TMlpANN ) histNnT ->Fill( reader->EvaluateMVA( "TMlpANN method" ) );
+ if (Use_BDT ) histBdt ->Fill( reader->EvaluateMVA( "BDT method" ) );
+ if (Use_BDTD ) histBdtD->Fill( reader->EvaluateMVA( "BDTD method" ) );
+ if (Use_RuleFit ) histRf ->Fill( reader->EvaluateMVA( "RuleFit method" ) );
+ }
+ cout << "--- end of event loop" << endl;
+ // get elapsed time
+ if (Use_Cuts) cout << "--- efficiency for cut method: " << double(nSelCuts)/theTree->GetEntries()
+ << " (for a required signal efficiency of " << effS << ")" << endl;
+
+ //
+ // write histograms
+ //
+ TFile *target = new TFile( "TMVApp.root","RECREATE" );
+ if (Use_Likelihood ) histLk ->Write();
+ if (Use_LikelihoodD ) histLkD ->Write();
+ if (Use_PDERS ) histPD ->Write();
+ if (Use_PDERSD ) histPDD ->Write();
+ if (Use_HMatrix ) histHm ->Write();
+ if (Use_Fisher ) histFi ->Write();
+ if (Use_MLP ) histNn ->Write();
+ if (Use_CFMlpANN ) histNnC ->Write();
+ if (Use_TMlpANN ) histNnT ->Write();
+ if (Use_BDT ) histBdt ->Write();
+ if (Use_BDTD ) histBdtD->Write();
+ if (Use_RuleFit ) histRf ->Write();
+ target->Close();
+
+ cout << "--- created root file: \"TMVApp.root\" containing the MVA output histograms" << endl;
+
+ delete reader;
+
+ cout << "==> TMVApplication is done!" << endl << endl;
+}
diff --git a/tmva/test/annconvergencetest.C b/tmva/test/annconvergencetest.C
new file mode 100644
index 00000000000..a97402a5c2f
--- /dev/null
+++ b/tmva/test/annconvergencetest.C
@@ -0,0 +1,67 @@
+#include "tmvaglob.C"
+
+// this macro serves to assess the convergence of the MLP ANN.
+// It compares the error estimator for the training and testing samples.
+// If overtraining occurred, the estimator for the training sample should
+// monotoneously decrease, while the estimator of the testing sample should
+// show a minimum after which it increases.
+
+// input: - Input file (result from TMVA),
+// - use of TMVA plotting TStyle
+void annconvergencetest( TString fin = "TMVA.root", bool useTMVAStyle=kTRUE )
+{
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TMVAGlob::OpenFile( fin );
+
+ TDirectory * dir = (TDirectory*)gDirectory->Get("MLP");
+ if (dir==0) {
+ cout << "Could not locate directory MLP in file " << fin << endl;
+ return;
+ }
+ dir->cd();
+
+ TCanvas* c = new TCanvas( "MLPConvergenceTest", "MLP Convergence Test", 150, 0, 600, 580*0.8 );
+
+ Double_t m1 = estimatorHistTrain->GetMaximum();
+ Double_t m2 = estimatorHistTest ->GetMaximum();
+ Double_t max = TMath::Max( m1, m2 );
+ m1 = estimatorHistTrain->GetMinimum();
+ m2 = estimatorHistTest ->GetMinimum();
+ Double_t min = TMath::Min( m1, m2 );
+ estimatorHistTrain->SetMaximum( max + 0.1*(max - min) );
+ estimatorHistTrain->SetMinimum( min - 0.1*(max - min) );
+ estimatorHistTrain->SetLineColor( 2 );
+ estimatorHistTrain->SetLineWidth( 2 );
+ estimatorHistTrain->SetTitle( TString("MLP Convergence test") );
+
+ estimatorHistTest->SetLineColor( 4 );
+ estimatorHistTest->SetLineWidth( 2 );
+
+ estimatorHistTrain->GetXaxis()->SetTitle( "Epochs" );
+ estimatorHistTrain->GetYaxis()->SetTitle( "Estimator" );
+ estimatorHistTrain->GetXaxis()->SetTitleOffset( 1.20 );
+ estimatorHistTrain->GetYaxis()->SetTitleOffset( 1.65 );
+
+ estimatorHistTrain->Draw();
+ estimatorHistTest ->Draw("same");
+
+ // need a legend
+ TLegend *legend= new TLegend( 1 - gPad->GetRightMargin() - 0.45, 1-gPad->GetTopMargin() - 0.20,
+ 1 - gPad->GetRightMargin() - 0.05, 1-gPad->GetTopMargin() - 0.05 );
+
+ legend->AddEntry(estimatorHistTrain,"Training Sample","l");
+ legend->AddEntry(estimatorHistTest,"Test sample","l");
+ legend->Draw("same");
+ legend->SetBorderSize(1);
+ legend->SetMargin( 0.3 );
+
+ c->cd();
+ TMVAGlob::plot_logo(); // don't understand why this doesn't work ... :-(
+ c->Update();
+
+ TString fname = "plots/annconvergencetest";
+ TMVAGlob::imgconv( c, fname );
+}
diff --git a/tmva/test/compareanapp.C b/tmva/test/compareanapp.C
new file mode 100644
index 00000000000..7f2029391b4
--- /dev/null
+++ b/tmva/test/compareanapp.C
@@ -0,0 +1,155 @@
+#include "tmvaglob.C"
+
+void compareanapp( TString finAn = "TMVA.root", TString finApp = "TMVApp.root", bool useTMVAStyle=kTRUE )
+{
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // switches
+ const Bool_t Draw_CFANN_Logy = kFALSE;
+ const Bool_t Save_Images = kTRUE;
+
+ cout << "Reading file: " << finApp << endl;
+ TFile *fileApp = new TFile( finApp );
+
+ // define Canvas layout here!
+ Int_t xPad = 1; // no of plots in x
+ Int_t yPad = 1; // no of plots in y
+ Int_t noPad = xPad * yPad ;
+ const Int_t width = 600; // size of canvas
+
+ // this defines how many canvases we need
+ const Int_t noCanvas = 10;
+ TCanvas **c = new TCanvas*[noCanvas];
+ for (Int_t ic=0; ic<noCanvas; ic++) c[ic] = 0;
+
+ // counter variables
+ Int_t countCanvas = 0;
+ Int_t countPad = 1;
+
+ // list of existing MVAs
+ const Int_t nmva = 16;
+ TString prefix = "";
+
+ // list of existing MVAs
+ const Int_t nmva = 16;
+ TString prefix = "";
+ TString mvaName[nmva] = { "MVA_Likelihood",
+ "MVA_LikelihoodD",
+ "MVA_Fisher",
+ "MVA_Fisher_Fi",
+ "MVA_Fisher_Ma",
+ "MVA_CFMlpANN",
+ "MVA_TMlpANN",
+ "MVA_HMatrix",
+ "MVA_PDERS" ,
+ "MVA_BDT",
+ "MVA_BDTGini",
+ "MVA_BDTMisCl",
+ "MVA_BDTStatSig",
+ "MVA_BDTCE",
+ "MVA_MLP",
+ "MVA_RuleFit" };
+ char fname[200];
+
+ // loop over MVAs
+ for (Int_t imva=0; imva<nmva; imva++) {
+
+ // retrieve corresponding signal and background histograms
+ TH1* all = (TH1*)gDirectory->Get( prefix + mvaName[imva] );
+
+ // only a single type (from TMVApplication.cpp)
+ if (NULL != all) {
+
+ TString hname = all->GetName();
+ cout << "--- Found (all) histogram: " << hname << endl;
+
+ // create new canvas
+ if ((c[countCanvas]==NULL) || (countPad>noPad)) {
+ cout << "--- Book canvas no: " << countCanvas << endl;
+ char cn[20];
+ sprintf( cn, "canvas%d", countCanvas+1 );
+ c[countCanvas] = new TCanvas( cn, "MVA Output Variables",
+ countCanvas*50, countCanvas*20, width, width*0.8 );
+ // style
+ c[countCanvas]->SetBorderMode(0);
+ c[countCanvas]->SetFillColor(10);
+
+ c[countCanvas]->Divide(xPad,yPad);
+ countPad = 1;
+ }
+
+ // chop off useless stuff
+ TString title(all->GetTitle());
+ title.ReplaceAll(prefix, "");
+ all->SetTitle( TString("MVA output for method: ") + title );
+
+ // set the histogram style
+ all->SetLineWidth( 2 );
+ all->SetLineColor( 1 );
+
+ // normalize 'all'
+ all->Scale( 1.0/all->GetSumOfWeights() );
+
+ // frame limits (choose judicuous x range)
+ Float_t nrms = 4;
+ cout << "--- mean and RMS: " << all->GetMean() << ", " << all->GetRMS() << endl;
+ Float_t xmin = TMath::Max( all->GetMean() - nrms*all->GetRMS(),
+ all->GetXaxis()->GetXmin() );
+ Float_t xmax = TMath::Min( all->GetMean() + nrms*all->GetRMS(),
+ all->GetXaxis()->GetXmax() );
+ Float_t ymin = 0;
+ Float_t ymax = all->GetMaximum()*1.2 ;
+
+ if (Draw_CFANN_Logy && mvaName[imva] == "CFANN") ymin = 0.01;
+
+ // build a frame
+ Int_t nb = 500;
+ TH2F* frame = new TH2F( TString("frame") + all->GetName(), all->GetTitle(),
+ nb, xmin, xmax, nb, ymin, ymax );
+ frame->GetXaxis()->SetTitle(title);
+ frame->GetYaxis()->SetTitle("Normalized");
+ TMVAGlob::SetFrameStyle( frame );
+
+ // eventually: draw the frame
+ frame->Draw();
+
+ if (Draw_CFANN_Logy && mvaName[imva] == "CFANN") gPad->SetLogy();
+
+ // add the reference histograms
+ // delete fileApp;
+ TFile *fileAn = new TFile( finAn );
+ TH1* sig = (TH1*)gDirectory->Get( prefix + mvaName[imva] + "_S" );
+ TH1* bgd = (TH1*)gDirectory->Get( prefix + mvaName[imva] + "_B" );
+ if (NULL != sig && NULL != bgd) {
+ // set the histogram style
+ TMVAGlob::SetSignalAndBackgroundStyle( sig, bgd );
+ // overlay signal and background histograms
+ sig->Scale( 1.0/sig->GetSumOfWeights() );
+ bgd->Scale( 1.0/bgd->GetSumOfWeights() );
+ sig->Draw("samehist");
+ bgd->Draw("samehist");
+ }
+
+ // overlay signal and background histograms
+ all->Draw("samehist");
+
+ // reopen TMVApplication output
+ TFile *fileApp = new TFile( finApp );
+
+ // redraw axes
+ frame->Draw("sameaxis");
+
+ // save canvas to file
+ c[countCanvas]->cd(countPad);
+ countPad++;
+ if (countPad > noPad) {
+ TMVAGlob::plot_logo();
+ c[countCanvas]->Update();
+ sprintf( fname, "plots/compareanapp_c%i", countCanvas+1 );
+ if (Save_Images) TMVAGlob::imgconv( c[countCanvas], &fname[0] );
+ countCanvas++;
+ }
+ }
+ }
+}
diff --git a/tmva/test/correlations.C b/tmva/test/correlations.C
index a97b2a9fece..b97ee50f21a 100644
--- a/tmva/test/correlations.C
+++ b/tmva/test/correlations.C
@@ -1,24 +1,30 @@
#include "tmvaglob.C"
-void correlations( TString fin = "TMVA.root", Bool_t greyScale = kFALSE )
+// this macro plots the correlation matrix of the various input
+// variables used in TMVA (e.g. running TMVAnalysis.C). Signal and
+// Background are plotted separately
+
+// input: - Input file (result from TMVA),
+// - use of colors or grey scale
+// - use of TMVA plotting TStyle
+void correlations( TString fin = "TMVA.root", Bool_t greyScale = kFALSE, Bool_t useTMVAStyle = kTRUE )
{
- gROOT->SetStyle("Plain");
- gStyle->SetOptStat(0);
- TList * loc = gROOT->GetListOfCanvases();
- TListIter itc(loc);
- TObject *o(0);
- while( (o = itc()) ) delete o;
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* file = TMVAGlob::OpenFile( fin );
// signal and background
const TString hName[2] = { "CorrelationMatrixS", "CorrelationMatrixB" };
const Int_t width = 600;
for (Int_t ic=0; ic<2; ic++) {
- TFile *file = new TFile( fin );
- TCanvas* c = new TCanvas( hName[ic], Form("Correlations between MVA input variables (%s)", (ic==0?"signal":"background")),
- ic*(width+5)+300, 0, width, width );
+ TCanvas* c = new TCanvas( hName[ic],
+ Form("Correlations between MVA input variables (%s)", (ic==0?"signal":"background")),
+ ic*(width+5)+200, 0, width, width );
Float_t newMargin1 = 0.13;
- Float_t newMargin2 = 0.18;
+ Float_t newMargin2 = 0.25;
gPad->SetGrid();
gPad->SetTicks();
@@ -32,7 +38,7 @@ void correlations( TString fin = "TMVA.root", Bool_t greyScale = kFALSE )
h2->SetMarkerSize( 1.5 );
h2->SetMarkerColor( 0 );
- Float_t labelSize = 0.050;
+ Float_t labelSize = 0.040;
h2->GetXaxis()->SetLabelSize( labelSize );
h2->GetYaxis()->SetLabelSize( labelSize );
h2->LabelsOption( "d" );
@@ -49,14 +55,13 @@ void correlations( TString fin = "TMVA.root", Bool_t greyScale = kFALSE )
h2->Draw("textsame"); // add text
// add comment
- TText* t = new TText( 0.31, 0.88, "absolute values for correlation coefficients given in %" );
+ TText* t = new TText( 0.53, 0.88, "linear correlation coefficients in %" );
t->SetNDC();
t->SetTextSize( 0.026 );
t->AppendPad();
- c->Modified();
-
- TMVAGlob::plot_logo( 0.85 );
+ TMVAGlob::plot_logo( );
+ c->Update();
TString fname = "plots/";
fname += hName[ic];
diff --git a/tmva/test/correlationscatters.C b/tmva/test/correlationscatters.C
new file mode 100644
index 00000000000..65b82b3b0b9
--- /dev/null
+++ b/tmva/test/correlationscatters.C
@@ -0,0 +1,160 @@
+#include "tmvaglob.C"
+
+// this macro plots the correlations (as scatter plots) of
+// the various input variable combinations used in TMVA (e.g. running
+// TMVAnalysis.C). Signal and Background are plotted separately
+
+const TString extensions[TMVAGlob::kNumOfMethods] = { "",
+ "_decorr",
+ "_PCA" };
+
+// input: - Input file (result from TMVA),
+// - normal/decorrelated/PCA
+// - use of TMVA plotting TStyle
+void correlationscatters( TString fin = "TMVA.root", TMVAGlob::TypeOfPlot type = TMVAGlob::kNormal,
+ bool useTMVAStyle = kTRUE )
+{
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* file = TMVAGlob::OpenFile( fin );
+
+ TString dirName = "CorrelationPlots" + extensions[type];
+
+ TDirectory* dir = (TDirectory*)gDirectory->Get( dirName );
+ if (dir==0) {
+ cout << "Could not locate directory: " << dirName << " in file " << fin << endl;
+ return;
+ }
+ dir->cd();
+
+ TListIter keyIt(dir->GetListOfKeys());
+ Int_t noPlots = 0;
+ TKey * key = 0;
+ // how many plots are in the directory?
+ Int_t noPlots = ((dir->GetListOfKeys())->GetEntries()) / 2;
+ cout << "--- Found: " << noPlots << " plots in directory: " << dirName << endl;
+
+ // define Canvas layout here!
+ // default setting
+ Int_t xPad; // no of plots in x
+ Int_t yPad; // no of plots in y
+ Int_t width; // size of canvas
+ Int_t height;
+ switch (noPlots) {
+ case 1:
+ xPad = 1; yPad = 1; width = 500; height = width; break;
+ case 2:
+ xPad = 2; yPad = 1; width = 600; height = 0.7*width; break;
+ case 3:
+ xPad = 3; yPad = 1; width = 800; height = 0.5*width; break;
+ case 4:
+ xPad = 2; yPad = 2; width = 600; height = width; break;
+ default:
+ xPad = 3; yPad = 2; width = 800; height = 0.7*width; break;
+ }
+ Int_t noPad = xPad * yPad ;
+
+ // this defines how many canvases we need
+ const Int_t noCanvas = 1 + (Int_t)(noPlots/noPad);
+ TCanvas **c = new TCanvas*[noCanvas];
+ for (Int_t ic=0; ic<noCanvas; ic++) c[ic] = 0;
+
+ cout << "--- Found: " << noPlots << " plots; "
+ << "will produce: " << noCanvas << " canvas" << endl;
+
+ // counter variables
+ Int_t countCanvas = 0;
+ Int_t countPad = 1;
+
+ // loop over all objects in "input_variables" directory
+ TString thename[2] = { "_sig", "_bgd" };
+ for (UInt_t itype = 0; itype < 2; itype++) {
+
+ TIter next(gDirectory->GetListOfKeys());
+ TKey* key = 0;
+
+ while ((key = (TKey*)next())) {
+
+ if (key->GetCycle() != 1) continue;
+
+ // make sure, that we only look at histograms
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH1")) continue;
+ TH1 *scat = (TH1*)key->ReadObj();
+ TString hname= scat->GetName();
+
+ // check for all signal histograms
+ if (hname.EndsWith( thename[itype] + extensions[type] ) &&
+ hname.BeginsWith( "scat_" )) { // found a new signal plot
+
+ cout << hname << endl;
+
+ // create new canvas
+ if ((c[countCanvas]==NULL) || (countPad>noPad)) {
+ cout << "--- Book canvas no: " << countCanvas << endl;
+ char cn[20];
+ sprintf( cn, "canvas%d", countCanvas+1 );
+ c[countCanvas] = new TCanvas( cn, "Correlation Profiles",
+ countCanvas*50+200, countCanvas*20, width, height );
+ // style
+ c[countCanvas]->SetBorderMode(0);
+ c[countCanvas]->SetFillColor(0);
+
+ c[countCanvas]->Divide(xPad,yPad);
+ countPad = 1;
+ }
+
+ // save canvas to file
+ c[countCanvas]->cd(countPad);
+ countPad++;
+
+ // find the corredponding backgrouns histo
+ TString bgname = hname;
+ bgname.ReplaceAll("scat_","prof_");
+ TH1 *prof = (TH1*)gDirectory->Get(bgname);
+ if (prof == NULL) {
+ cout << "ERROR!!! couldn't find backgroung histo for" << hname << endl;
+ exit;
+ }
+ // this is set but not stored during plot creation in MVA_Factory
+ TMVAGlob::SetSignalAndBackgroundStyle( scat, prof );
+
+ // chop off "signal"
+ TMVAGlob::SetFrameStyle( scat, 1.2 );
+
+ // finally plot and overlay
+ Float_t sc = 1.1;
+ if (countPad==2) sc = 1.3;
+ scat->SetMarkerColor( 4);
+ scat->Draw();
+ prof->SetMarkerColor( 2 );
+ prof->SetMarkerSize( 0.2 );
+ prof->SetLineColor( 2 );
+ prof->SetLineWidth( 1 );
+ prof->SetFillStyle( 3002 );
+ prof->SetFillColor( 46 );
+ prof->Draw("samee1");
+
+ // redraw axes
+ scat->Draw("sameaxis");
+
+ // save canvas to file
+ if (countPad > noPad) {
+ c[countCanvas]->Update();
+
+ TString fname = Form( "plots/correlationscatter_%s_c%i", extensions[type].Data(), countCanvas+1 );
+ TMVAGlob::imgconv( c[countCanvas], &fname[0] );
+ countCanvas++;
+ }
+ }
+ }
+ }
+ if (countPad <= noPad) {
+ c[countCanvas]->Update();
+ TString fname = Form( "plots/correlationscatter_%s_c%i", extensions[type].Data(), countCanvas+1 );
+ TMVAGlob::imgconv( c[countCanvas], &fname[0] );
+ }
+
+}
diff --git a/tmva/test/efficiencies.C b/tmva/test/efficiencies.C
index 49a615d5b0b..1a3c99f3b9b 100644
--- a/tmva/test/efficiencies.C
+++ b/tmva/test/efficiencies.C
@@ -1,54 +1,17 @@
#include "tmvaglob.C"
-void efficiencies( TString fin = "TMVA.root", Int_t type = 2 )
-{
- // argument: type = 1 --> plot efficiency(B) versus eff(S)
- // type = 2 --> plot rejection (B) versus efficiency (S)
-
- gROOT->SetStyle("Plain");
- gStyle->SetOptStat(0);
- TList * loc = gROOT->GetListOfCanvases();
- TListIter itc(loc);
- TObject *o(0);
- while( (o = itc()) ) delete o;
- cout << "Reading file: " << fin << endl;
- TFile *file = new TFile( fin );
- // check if multi-cut MVA or only one set of MVAs
- Bool_t multiMVA=kFALSE;
- TIter nextDir(file->GetListOfKeys());
- TKey *key;
- // loop over all directories and check if
- // one contains the key word 'multicutMVA'
- while (key = (TKey*)nextDir()) {
- TClass *cl = gROOT->GetClass(key->GetClassName());
- if (!cl->InheritsFrom("TDirectory")) continue;
- TDirectory *d = (TDirectory*)key->ReadObj();
- TString path(d->GetPath());
- if ((TString(d->GetPath())).Contains("multicutMVA")){
- multiMVA=kTRUE;
- plot_efficiencies(fin,type,d);
- }
- }
- plot_efficiencies(fin,type,gDirectory);
-
-}
-
-void plot_efficiencies( TString fin = "TMVA.root", Int_t type = 2 , TDirectory* BinDir)
+void plot_efficiencies( TFile* file, Int_t type = 2 , TDirectory* BinDir)
{
- // argument: type = 1 --> plot efficiency(B) versus eff(S)
- // type = 2 --> plot rejection (B) versus efficiency (S)
+ // input: - Input file (result from TMVA),
+ // - type = 1 --> plot efficiency(B) versus eff(S)
+ // = 2 --> plot rejection (B) versus efficiency (S)
Bool_t __PRINT_LOGO__ = kTRUE;
Bool_t __SAVE_IMAGE__ = kTRUE;
- cout <<"Bindir="<<BinDir->GetName()<<endl;
- gROOT->Reset();
- gROOT->SetStyle("Plain");
- gStyle->SetOptStat(0);
-
- TFile *file = new TFile( fin );
+ // cout << "Bindir=" << BinDir->GetName() << endl;
// the coordinates
Float_t x1 = 0;
@@ -61,13 +24,13 @@ void plot_efficiencies( TString fin = "TMVA.root", Int_t type = 2 , TDirectory*
Float_t z = y1;
y1 = 1 - y2;
y2 = 1 - z;
- cout << "--- type==2: plot background rejection versus signal efficiency" << endl;
+ // cout << "--- type==2: plot background rejection versus signal efficiency" << endl;
+ }
+ else {
+ // cout << "--- type==1: plot background efficiency versus signal efficiency" << endl;
}
- else
- cout << "--- type==1: plot background efficiency versus signal efficiency" << endl;
-
// create canvas
- TCanvas* c = new TCanvas( "c", "the canvas", 300, 0, 650, 500 );
+ TCanvas* c = new TCanvas( "c", "the canvas", 200, 0, 650, 500 );
c->SetBorderMode(0);
c->SetFillColor(10);
@@ -77,7 +40,6 @@ void plot_efficiencies( TString fin = "TMVA.root", Int_t type = 2 , TDirectory*
// legend
Float_t x0L = 0.107, y0H = 0.899;
- //Float_t dxL = 0.557-x0L, dyH = 0.22;
Float_t dxL = 0.457-x0L, dyH = 0.22;
if (type == 2) {
x0L = 0.15;
@@ -88,6 +50,7 @@ void plot_efficiencies( TString fin = "TMVA.root", Int_t type = 2 , TDirectory*
legend->SetTextSize( 0.05 );
legend->SetHeader( "MVA Method:" );
legend->SetMargin( 0.4 );
+ legend->SetFillColor(0);
TString xtit = "Signal efficiency";
TString ytit = "Background efficiency";
@@ -99,6 +62,7 @@ void plot_efficiencies( TString fin = "TMVA.root", Int_t type = 2 , TDirectory*
ftit += (BinDir->GetTitle());
}
// draw empty frame
+ if(gROOT->FindObject("frame")!=0) gROOT->FindObject("frame")->Delete();
TH2F* frame = new TH2F( "frame", ftit, 500, x1, x2, 500, y1, y2 );
frame->GetXaxis()->SetTitle( xtit );
frame->GetYaxis()->SetTitle( ytit );
@@ -111,25 +75,44 @@ void plot_efficiencies( TString fin = "TMVA.root", Int_t type = 2 , TDirectory*
TIter next(file->GetListOfKeys());
TKey *key;
- TString hName = "effBvsS";
- if (type == 2) hName = "rejBvsS";
+ TString hNameRef = "effBvsS";
+ if (type == 2) hNameRef = "rejBvsS";
+
+ TList hists;
// loop over all histograms with that name
while (key = (TKey*)next()) {
TClass *cl = gROOT->GetClass(key->GetClassName());
if (!cl->InheritsFrom("TH1")) continue;
TH1 *h = (TH1*)key->ReadObj();
- if ((TString(h->GetName()).Contains( hName))&&
- (TString(h->GetName()).Contains( "MVA_" ))){
+ TString hname = h->GetName();
+
+ if (hname.Contains( hNameRef ) && hname.BeginsWith( "MVA_" )) {
h->SetLineWidth(3);
h->SetLineColor(color);
color++; if (color == 5 || color == 10 || color == 11) color++;
- legend->AddEntry(h,TString(h->GetTitle()).ReplaceAll("MVA_",""),"l");
h->Draw("csame");
+ hists.Add(h);
nmva++;
}
}
-
+
+ while (hists.GetSize()) {
+ TListIter hIt(&hists);
+ TH1* hist(0);
+ Double_t largestInt=0;
+ TH1* histWithLargestInt(0);
+ while ((hist = (TH1*)hIt())!=0) {
+ Double_t integral = hist->Integral(1,hist->FindBin(0.9999));
+ if (integral>largestInt) {
+ largestInt = integral;
+ histWithLargestInt = hist;
+ }
+ }
+ legend->AddEntry(histWithLargestInt,TString(histWithLargestInt->GetTitle()).ReplaceAll("MVA_",""),"l");
+ hists.Remove(histWithLargestInt);
+ }
+
// rescale legend box size
// current box size has been tuned for 3 MVAs + 1 title
if (type == 1) {
@@ -142,8 +125,7 @@ void plot_efficiencies( TString fin = "TMVA.root", Int_t type = 2 , TDirectory*
}
// redraw axes
- frame->Draw("sameaxis");
-
+ frame->Draw("sameaxis");
legend->Draw("same");
// ============================================================
@@ -154,14 +136,46 @@ void plot_efficiencies( TString fin = "TMVA.root", Int_t type = 2 , TDirectory*
c->Update();
- TString fname = "plots/" + hName;
+ TString fname = "plots/" + hNameRef;
if (TString(BinDir->GetName()).Contains("multicut")){
TString fprepend(BinDir->GetName());
fprepend.ReplaceAll("multicutMVA_","");
- TString fname = "plots/" + fprepend + "_" + hName;
- }else {
- String fname = "plots/" + hName;
+ fname = "plots/" + fprepend + "_" + hNameRef;
}
if (__SAVE_IMAGE__) TMVAGlob::imgconv( c, fname );
+ return;
+}
+
+void efficiencies( TString fin = "TMVA.root", Int_t type = 2, Bool_t useTMVAStyle = kTRUE )
+{
+ // argument: type = 1 --> plot efficiency(B) versus eff(S)
+ // type = 2 --> plot rejection (B) versus efficiency (S)
+
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* file = TMVAGlob::OpenFile( fin );
+
+ // check if multi-cut MVA or only one set of MVAs
+ Bool_t multiMVA=kFALSE;
+ TIter nextDir(file->GetListOfKeys());
+ TKey *key;
+ // loop over all directories and check if
+ // one contains the key word 'multicutMVA'
+ while ((key = (TKey*)nextDir())) {
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TDirectory")) continue;
+ TDirectory *d = (TDirectory*)key->ReadObj();
+ TString path(d->GetPath());
+ if ((TString(d->GetPath())).Contains("multicutMVA")){
+ multiMVA=kTRUE;
+ plot_efficiencies( file, type, d );
+ }
+ }
+ plot_efficiencies( file, type, gDirectory );
+
+ return;
}
+
diff --git a/tmva/test/likelihoodrefs.C b/tmva/test/likelihoodrefs.C
new file mode 100644
index 00000000000..bf8b59a0820
--- /dev/null
+++ b/tmva/test/likelihoodrefs.C
@@ -0,0 +1,181 @@
+#include <vector>
+#include <string>
+#include "tmvaglob.C"
+
+
+// this macro plots the reference distribuions for the Likelihood
+// methods for the various input variables used in TMVA (e.g. running
+// TMVAnalysis.C). Signal and Background are plotted separately
+
+
+// input: - Input file (result from TMVA),
+// - use of TMVA plotting TStyle
+void likelihoodrefs( TString fin = "TMVA.root", Bool_t useTMVAStyle = kTRUE )
+{
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* file = TMVAGlob::OpenFile( fin );
+
+ file->cd("Likelihood");
+ TDirectory *current_sourcedir = gDirectory;
+ Int_t color=1;
+ TIter next(current_sourcedir->GetListOfKeys());
+ TKey *key;
+ TLegend *legS = new TLegend(0.14,0.7,0.87,0.87);
+ TLegend *legB = new TLegend(0.14,0.7,0.87,0.87);
+
+ Bool_t newCanvas = kTRUE;
+
+ const UInt_t maxCanvas = 200;
+ TCanvas** c = new TCanvas*[maxCanvas];
+ Int_t width = 500;
+ Int_t height = 500;
+
+ // avoid duplicated printing
+ std::vector<std::string> hasBeenUsed;
+
+ UInt_t ic = -1;
+
+ while ((key = (TKey*)next())) {
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH1")) continue;
+ TH1 *h = (TH1*)key->ReadObj();
+ TH1F *b( 0 );
+ TString hname( h->GetName() );
+
+ // avoid duplicated plotting
+ Bool_t found = kFALSE;
+ for (UInt_t j = 0; j < hasBeenUsed.size(); j++) {
+ if (hasBeenUsed[j] == hname.Data()) found = kTRUE;
+ }
+ if (!found) {
+
+ // draw original plots
+ if (hname.EndsWith("_sig")) {
+
+ if (newCanvas) {
+ char cn[20];
+ sprintf( cn, "canvas%d", ic+1 );
+ ++ic;
+ TString n = hname;
+ c[ic] = new TCanvas( cn, Form( "Likelihood reference for variable: %s",
+ (n.ReplaceAll("_sig","")).Data() ),
+ ic*50+200, ic*20, width, height );
+ c[ic]->Divide(2,2);
+ newCanvas = kFALSE;
+ }
+
+ // signal
+ h->SetMaximum(h->GetMaximum()*1.1);
+ color = 4;
+ c[ic]->cd(1);
+ TString plotname = hname;
+ legS->Clear();
+ legS->SetBorderSize(1);
+ h->SetMarkerColor(color);
+ h->SetMarkerSize( 0.7 );
+ h->SetMarkerStyle( 20 );
+ h->SetLineWidth(1);
+ h->SetLineColor(color);
+ color++;
+ legS->AddEntry(h,"Input data (signal)","p");
+ h->Draw("e1");
+
+ // background
+ TString bname( hname );
+ b = (TH1F*)gDirectory->Get( bname.ReplaceAll("_sig","_bgd") );
+ c[ic]->cd(3);
+ color = 2;
+ legB->Clear();
+ legB->SetBorderSize(1);
+ b->SetMaximum(b->GetMaximum()*1.1);
+ b->SetLineWidth(1);
+ b->SetLineColor(color);
+ b->SetMarkerColor(color);
+ b->SetMarkerSize( 0.7 );
+ b->SetMarkerStyle( 20 );
+ legB->AddEntry(b,"Input data (backgr.)","p");
+ b->Draw("e1");
+
+ // register
+ hasBeenUsed.push_back( bname.Data() );
+
+ // the smooth histograms
+ TString hsmooth = hname + "_smooth";
+ h = (TH1F*)gDirectory->Get( hsmooth );
+ if (h == 0) {
+ cout << "ERROR in likelihoodrefs.C: unknown histogram: " << hsmooth << endl;
+ return;
+ }
+ b = (TH1F*)gDirectory->Get( hsmooth.ReplaceAll("_sig","_bgd") );
+
+ color = 1;
+ c[ic]->cd(1);
+ h->SetLineWidth(2);
+ h->SetLineColor(color);
+ h->SetMarkerColor(color);
+ color++;
+ legS->AddEntry(h,"Smoothed histogram (signal)","l");
+ h->Draw("histsame");
+
+ color = 1;
+ c[ic]->cd(3);
+ b->SetLineWidth(2);
+ b->Draw("histsame");
+ legB->AddEntry(b,"Smoothed histogram (backgr.)","l");
+
+ hasBeenUsed.push_back( hname.Data() );
+
+ // the splines
+ for (int i=0; i<= 5; i++) {
+ TString hspline = hname + Form( "_smooth_hist_from_spline%i", i );
+ h = (TH1F*)gDirectory->Get( hspline );
+ if (h) {
+ b = (TH1F*)gDirectory->Get( hspline.ReplaceAll("_sig","_bgd") );
+ break;
+ }
+ }
+ if (h == 0 || b == 0) {
+ cout << "--- likelihoodrefs.C: did not find spline for histogram: " << hname.Data() << endl;
+ }
+ else {
+
+ h->SetMaximum(h->GetMaximum()*1.5);
+ color = 4;
+ c[ic]->cd(2);
+ h->SetLineWidth(2);
+ h->SetLineColor(color);
+ legS->AddEntry(h,"Splined PDF (norm. signal)","l");
+ h->Draw("hist");
+ legS->Draw();
+
+ b->SetMaximum(b->GetMaximum()*1.5);
+ color = 2;
+ c[ic]->cd(4);
+ b->SetLineColor(color);
+ b->SetLineWidth(2);
+ legB->AddEntry(b,"Splined PDF (norm. backgr.)","l");
+ b->Draw("hist");
+
+ // draw the legends
+ legB->Draw();
+
+ hasBeenUsed.push_back( hname.Data() );
+ }
+
+ c[ic]->Update();
+
+ // write to file
+ TString fname = Form( "plots/likelihoodrefs_c%i", ic+1 );
+ TMVAGlob::imgconv( c[ic], fname );
+ // c[ic]->Update();
+
+ newCanvas = kTRUE;
+ hasBeenUsed.push_back( hname.Data() );
+ }
+ }
+ }
+}
+
diff --git a/tmva/test/line-small.png b/tmva/test/line-small.png
new file mode 100644
index 0000000000000000000000000000000000000000..9e9d90045a7a93206e94b7f1f534a3e3120d82bb
GIT binary patch
literal 294
zcmeAS@N?(olHy`uVBq!ia0y~yU@!t<4kiW$hKaHA&lngOSkfJR9T^xl_H+M9WMyDr
zxa;ZS7!u)r_5vf{VFezRgCG982QLcAnk{5-Y~SwMgBjjyzY46LQe-kMB==&a*M%iV
zciF|Ryz|R!<%T>p&TIQS&%c^=VvV}+WK)BS5y`^srlx`sV#2dhO&lZ3g!{IpFl`L;
z<(PJO%DGQ7iz4%am9mddIeN=v>4tB<AJ+&38)%;rImu<xx-rSua7{rlgZAMqr@f4F
zH-7S|3tM&a<SDbt2!_i>+{sbN2YtA8mYh7rG}}jKPv*HrGt_jZWgbzQB^MbJJm1y4
yG=giHwfB)Phc6k{Mg%Q0_dZ*)iN&|~E7KuG?q=^zqRk8p3=E#GelF{r5}E+a_;m;X
literal 0
HcmV?d00001
diff --git a/tmva/test/mutransform.C b/tmva/test/mutransform.C
new file mode 100644
index 00000000000..631a0a48db8
--- /dev/null
+++ b/tmva/test/mutransform.C
@@ -0,0 +1,137 @@
+#include "tmvaglob.C"
+
+// this macro plots the mu-transformation results of the MVA output
+// variables from the various MVA methods run in TMVA (e.g. running
+// TMVAnalysis.C). Since the mu-transform is flat for background, only
+// signal is shown. The more signal ispeaked towards one, the better is
+// the discrimination of the MVA method. See this reference for more
+// information (in French):
+// http://tel.archives-ouvertes.fr/documents/archives0/00/00/29/91/index_fr.html
+//
+// input: - Input file (result from TMVA),
+// - use log/lin scale
+// - use of TMVA plotting TStyle
+void mutransform( TString fin = "TMVA.root", Bool_t logy = kFALSE, Bool_t useTMVAStyle = kTRUE )
+{
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* file = TMVAGlob::OpenFile( fin );
+
+ // the coordinates
+ Float_t x1 = 0.0;
+ Float_t x2 = 1;
+ Float_t y1 = 0;
+ Float_t y2 = -1;
+ Float_t ys = 1.08;
+
+ enum { nskip = 1 };
+ TString hskip[nskip] = { "Variable" };
+
+ // loop over all histograms with that name
+ // search for maximum ordinate
+ TIter next(file->GetListOfKeys());
+ TKey * key;
+ Int_t nmva = 0;
+ TString hName = "muTransform_S"; // ignore background
+ while (key = (TKey*)next()) {
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH1")) continue;
+ TH1 *h = (TH1*)key->ReadObj();
+ Bool_t skip = !TString(h->GetName()).Contains( hName );
+ for (Int_t iskip=0; iskip<nskip; iskip++)
+ if (TString(h->GetName()).Contains( hskip[iskip] )) skip = kTRUE;
+ if (!skip) {
+ if (h->GetMaximum() > y2) y2 = h->GetMaximum()*ys;
+ nmva++;
+ }
+ }
+ if (y2 == -1) {
+ cout << "No mu-transforms found" << endl;
+ return;
+ }
+
+ // create canvas
+ TCanvas* c = new TCanvas( "c", "the canvas", 150, 0, 650, 500 );
+ c->SetBorderMode(0);
+ c->SetFillColor(10);
+
+ // global style settings
+ gPad->SetTicks();
+ if (logy) {
+ y1 = 0.1;
+ ys = 2.0;
+ gPad->SetLogy();
+ }
+
+ // legend
+ Float_t x0L = 0.140, y0H = 0.86;
+ Float_t dxL = 0.48, dyH = 0.22;
+ TLegend *legend = new TLegend( x0L, y0H-dyH, x0L+dxL, y0H );
+ legend->SetBorderSize(1);
+ legend->SetTextSize( 0.05 );
+ legend->SetHeader( "MVA Method:" );
+ legend->SetMargin( 0.4 );
+
+ TString xtit = "mu-transform";
+ TString ytit = "";
+ TString ftit = "Signal " + xtit;
+
+ cout << "--- set frame maximum to: " << y2 << endl;
+ next.Reset();
+
+ // rescale legend box size
+ // current box size has been tuned for 3 MVAs + 1 title
+ dyH *= (1.0 + Float_t(nmva - 3.0)/4.0);
+ legend->SetY1( y0H - dyH );
+
+ // draw empty frame
+ TH2F* frame = new TH2F( "frame", ftit, 500, x1, x2, 500, y1, y2 );
+ frame->GetXaxis()->SetTitle( xtit );
+ frame->GetYaxis()->SetTitle( ytit );
+ TMVAGlob::SetFrameStyle( frame, 1.0 );
+
+ frame->Draw();
+
+ // loop over all histograms with that name
+ // plot
+ Int_t color = 1;
+ while (key = (TKey*)next()) {
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH1")) continue;
+ TH1 *h = (TH1*)key->ReadObj();
+ Bool_t skip = !TString(h->GetName()).Contains( hName );
+ for (Int_t iskip=0; iskip<nskip; iskip++)
+ if (TString(h->GetName()).Contains( hskip[iskip] )) skip = kTRUE;
+ if (!skip) {
+ // signal or background ?
+ if (TString(h->GetName()).Contains( "_S" )) {
+ h->SetLineStyle( 1 );
+ h->SetLineWidth( 3 );
+ }
+ else {
+ h->SetLineStyle( 2 );
+ h->SetLineWidth( 3 );
+ }
+ h->SetLineColor(color);
+ color++;
+ TString tit = h->GetTitle();
+ tit.ReplaceAll( "mu-Transform", "" );
+ tit.ReplaceAll( "(S)", "" );
+ tit.ReplaceAll( ":", "" );
+ legend->AddEntry( h, tit, "l" );
+ h->Draw("same");
+ }
+ }
+
+ // redraw axes
+ frame->Draw("sameaxis");
+
+ legend->Draw("same");
+ c->Update();
+
+ TString fname = "plots/mutransform";
+ TMVAGlob::imgconv( c, fname );
+}
+
diff --git a/tmva/test/mvas.C b/tmva/test/mvas.C
index cf765df936c..ea36eaa12c5 100644
--- a/tmva/test/mvas.C
+++ b/tmva/test/mvas.C
@@ -1,26 +1,29 @@
#include "tmvaglob.C"
-void mvas( TString fin = "TMVA.root" )
+// this macro plots the resulting MVA distribution (Signal and
+// Background overlayed) of different MVA methods run in TMVA
+// (e.g. running TMVAnalysis.C).
+
+
+// input: - Input file (result from TMVA)
+// - use of TMVA plotting TStyle
+void mvas( TString fin = "TMVA.root", Bool_t useTMVAStyle = kTRUE )
{
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
// switches
const Bool_t Draw_CFANN_Logy = kFALSE;
const Bool_t Save_Images = kTRUE;
- gROOT->SetStyle("Plain");
- gStyle->SetOptStat(0);
- TList * loc = gROOT->GetListOfCanvases();
- TListIter itc(loc);
- TObject *o(0);
- while( (o = itc()) ) delete o;
-
- cout << "Reading file: " << fin << endl;
- TFile *file = new TFile( fin );
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* file = TMVAGlob::OpenFile( fin );
// define Canvas layout here!
Int_t xPad = 1; // no of plots in x
Int_t yPad = 1; // no of plots in y
Int_t noPad = xPad * yPad ;
- const Int_t width = 650; // size of canvas
+ const Int_t width = 600; // size of canvas
// this defines how many canvases we need
const Int_t noCanvas = 10;
@@ -32,31 +35,38 @@ void mvas( TString fin = "TMVA.root" )
Int_t countPad = 1;
// list of existing MVAs
- const Int_t nmva = 14;
- TString prefix = "";
- TString mvaName[nmva] = { "MVA_Likelihood",
- "MVA_LikelihoodD",
- "MVA_Fisher",
- "MVA_Fisher_Fi",
- "MVA_Fisher_Ma",
- "MVA_CFMlpANN",
- "MVA_TMlpANN",
- "MVA_HMatrix",
- "MVA_PDERS" ,
- "MVA_BDT",
- "MVA_BDTGini",
- "MVA_BDTMisCl",
- "MVA_BDTStatSig",
- "MVA_BDTCE" };
- char fname[200];
-
- // loop over MVAs
- for (Int_t imva=0; imva<nmva; imva++) {
+ const Int_t nveto = 1;
+ TString prefix = "MVA_";
+ TString suffixSig = "_S";
+ TString suffixBgd = "_B";
+ TString vetoNames[nveto] = { "muTransform" };
+
+ // search for the right histograms in full list of keys
+ TIter next(file->GetListOfKeys());
+ TKey *key;
+ char fname[200];
+ while ((key = (TKey*)next())) {
+
+ // make sure, that we only look at histograms
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH1")) continue;
+ TH1 *th1 = (TH1*)key->ReadObj();
+ TString hname= th1->GetName();
+
+ if (!hname.BeginsWith( prefix ) || !hname.EndsWith( suffixSig )) continue;
+
+ // check if histogram is vetoed
+ Bool_t found = kFALSE;
+ for (UInt_t iv=0; iv<nveto; iv++) if (hname.Contains( vetoNames[iv] )) found = kTRUE;
+ if (found) continue;
+
+ // remove the signal suffix
+ hname.ReplaceAll( suffixSig, "" );
// retrieve corresponding signal and background histograms
- TH1* sig = (TH1*)gDirectory->Get( prefix + mvaName[imva] + "_S" );
- TH1* bgd = (TH1*)gDirectory->Get( prefix + mvaName[imva] + "_B" );
- TH1* all = (TH1*)gDirectory->Get( prefix + mvaName[imva] );
+ TH1* sig = (TH1*)gDirectory->Get( hname + "_S" );
+ TH1* bgd = (TH1*)gDirectory->Get( hname + "_B" );
+ TH1* all = (TH1*)gDirectory->Get( hname );
// check that exist
if (NULL != sig && NULL != bgd) {
@@ -76,7 +86,7 @@ void mvas( TString fin = "TMVA.root" )
char cn[20];
sprintf( cn, "canvas%d", countCanvas+1 );
c[countCanvas] = new TCanvas( cn, Form("MVA Output Variables %s",title.Data()),
- countCanvas*50+300, countCanvas*20, width, width*0.8 );
+ countCanvas*50+200, countCanvas*20, width, width*0.78 );
// style
c[countCanvas]->SetBorderMode(0);
c[countCanvas]->SetFillColor(10);
@@ -85,7 +95,6 @@ void mvas( TString fin = "TMVA.root" )
countPad = 1;
}
-
// set the histogram style
TMVAGlob::SetSignalAndBackgroundStyle( sig, bgd );
@@ -134,13 +143,12 @@ void mvas( TString fin = "TMVA.root" )
legend->SetMargin( 0.3 );
}
- TMVAGlob::plot_logo();
-
// save canvas to file
c[countCanvas]->cd(countPad);
countPad++;
if (countPad > noPad) {
c[countCanvas]->Update();
+ TMVAGlob::plot_logo();
sprintf( fname, "plots/mva_c%i", countCanvas+1 );
if (Save_Images) TMVAGlob::imgconv( c[countCanvas], &fname[0] );
countCanvas++;
@@ -207,12 +215,11 @@ void mvas( TString fin = "TMVA.root" )
// redraw axes
frame->Draw("sameaxis");
- TMVAGlob::plot_logo();
-
// save canvas to file
c[countCanvas]->cd(countPad);
countPad++;
if (countPad > noPad) {
+ TMVAGlob::plot_logo();
c[countCanvas]->Update();
sprintf( fname, "plots/mva_all_c%i", countCanvas+1 );
if (Save_Images) TMVAGlob::imgconv( c[countCanvas], &fname[0] );
diff --git a/tmva/test/network.C b/tmva/test/network.C
new file mode 100644
index 00000000000..ce587738d3d
--- /dev/null
+++ b/tmva/test/network.C
@@ -0,0 +1,313 @@
+#include "tmvaglob.C"
+
+// this macro prints out a neural network generated by MethodMLP graphically
+// @author: Matt Jachowski, jachowski@stanford.edu
+
+// input: - Input file (result from TMVA),
+// - use of TMVA plotting TStyle
+void network( TString fin = "TMVA.root", Bool_t useTMVAStyle = kTRUE )
+{
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* file = TMVAGlob::OpenFile( fin );
+
+ TDirectory * dir = (TDirectory*)gDirectory->Get("MLP");
+ if (dir==0) {
+ cout << "Could not locate directory MLP in file " << fin << endl;
+ return;
+ }
+ dir->cd();
+ draw_network(dir);
+}
+
+void draw_network(TDirectory* d)
+{
+ Bool_t __PRINT_LOGO__ = kTRUE;
+
+ // create canvas
+ TCanvas* c = new TCanvas( "c", "Neural Network Layout", 100, 0, 1000, 650 );
+ c->SetBorderMode(0);
+ c->SetFillColor(10);
+
+ TIter next = d->GetListOfKeys();
+ TKey *key;
+ TString hName = "weights_hist";
+ Int_t numHists = 0;
+
+ // loop over all histograms with hName in name
+ while (key = (TKey*)next()) {
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH2F")) continue;
+ TH2F *h = (TH2F*)key->ReadObj();
+ if (TString(h->GetName()).Contains( hName ))
+ numHists++;
+ }
+
+ // loop over all histograms with hName in name again
+ next.Reset();
+ Double_t maxWeight = 0;
+
+ // find max weight
+ while (key = (TKey*)next()) {
+
+ //cout << "Title: " << key->GetTitle() << endl;
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH2F")) continue;
+
+ TH2F* h = (TH2F*)key->ReadObj();
+ if (TString(h->GetName()).Contains( hName )){
+
+ Int_t n1 = h->GetNbinsX();
+ Int_t n2 = h->GetNbinsY();
+ for (Int_t i = 0; i < n1; i++) {
+ for (Int_t j = 0; j < n2; j++) {
+ Double_t weight = TMath::Abs(h->GetBinContent(i+1, j+1));
+ if (maxWeight < weight) maxWeight = weight;
+ }
+ }
+ }
+ }
+
+ // draw network
+ next.Reset();
+ Int_t count = 0;
+ while (key = (TKey*)next()) {
+
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH2F")) continue;
+
+ TH2F* h = (TH2F*)key->ReadObj();
+ if (TString(h->GetName()).Contains( hName )){
+ draw_layer(c, h, count++, numHists+1, maxWeight);
+ }
+ }
+
+ draw_layer_labels(numHists+1);
+
+ // ============================================================
+ if (__PRINT_LOGO__) TMVAGlob::plot_logo();
+ // ============================================================
+
+ c->Update();
+
+ TString fname = "plots/network";
+ TMVAGlob::imgconv( c, fname );
+}
+
+void draw_layer_labels(Int_t nLayers)
+{
+ const Double_t LABEL_HEIGHT = 0.03;
+ const Double_t LABEL_WIDTH = 0.20;
+ Double_t effWidth = 0.8*(1.0-LABEL_WIDTH)/nLayers;
+ Double_t height = 0.8*LABEL_HEIGHT;
+ Double_t margY = LABEL_HEIGHT - height;
+
+ for (Int_t i = 0; i < nLayers; i++) {
+ TString label = Form("Layer %i", i);
+ Double_t cx = i*(1.0-LABEL_WIDTH)/nLayers+1.0/(2.0*nLayers)+LABEL_WIDTH;
+ Double_t x1 = cx-0.8*effWidth/2.0;
+ Double_t x2 = cx+0.8*effWidth/2.0;
+ Double_t y1 = margY;
+ Double_t y2 = margY + height;
+
+ TPaveLabel *p = new TPaveLabel(x1, y1, x2, y2, label+"", "br");
+ p->Draw();
+ }
+}
+
+void draw_input_labels(Int_t nInputs, Double_t* cy,
+ Double_t rad, Double_t layerWidth)
+{
+ const Double_t LABEL_HEIGHT = 0.03;
+ const Double_t LABEL_WIDTH = 0.20;
+ Double_t width = LABEL_WIDTH + (layerWidth-4*rad);
+ Double_t margX = 0.01;
+ Double_t effHeight = 0.8*LABEL_HEIGHT;
+
+ TString *varNames = get_var_names(nInputs);
+ TString input;
+
+ for (Int_t i = 0; i < nInputs; i++) {
+ if (i != nInputs-1) input = varNames[i];
+ else input = "bias";
+ Double_t x1 = margX;
+ Double_t x2 = margX + width;
+ Double_t y1 = cy[i] - effHeight;
+ Double_t y2 = cy[i] + effHeight;
+
+ TPaveLabel *p = new TPaveLabel(x1, y1, x2, y2, input+"", "br");
+ p->Draw();
+ if (i == nInputs-1) p->SetTextColor(9);
+ }
+
+ delete[] varNames;
+}
+
+TString* get_var_names(Int_t nVars)
+{
+ TString fname = "weights/MVAnalysis_MLP.weights.txt";
+ ifstream fin( fname );
+ if (!fin.good( )) { // file not found --> Error
+ cout << "Error opening " << fname << endl;
+ exit(1);
+ }
+
+ Int_t idummy;
+ Float_t fdummy;
+ TString dummy = "";
+
+ // file header with name
+ while (!dummy.Contains("#VAR")) fin >> dummy;
+ fin >> dummy >> dummy >> dummy; // the rest of header line
+
+ // number of variables
+ fin >> dummy >> idummy;
+ // at this point, we should have idummy == nVars
+
+ // variable mins and maxes
+ TString* vars = new TString[nVars];
+ for (Int_t i = 0; i < idummy; i++) fin >> vars[i] >> dummy >> dummy >> dummy;
+
+ fin.close();
+
+ return vars;
+}
+
+void draw_activation(TCanvas* c, Double_t cx, Double_t cy,
+ Double_t radx, Double_t rady, Int_t whichActivation)
+{
+ TImage *activation = NULL;
+
+ switch (whichActivation) {
+ case 0:
+ activation = TImage::Open("sigmoid-small.png");
+ break;
+ case 1:
+ activation = TImage::Open("line-small.png");
+ break;
+ default:
+ cout << "Activation index " << whichActivation << " is not known." << endl;
+ cout << "You messed up or you need to modify network.C to introduce a new "
+ << "activation function (and image) corresponding to this index" << endl;
+ }
+
+ if (activation == NULL) {
+ cout << "Could not create an image... exit" << endl;
+ return;
+ }
+
+ activation->SetConstRatio(kFALSE);
+
+ radx *= 0.7;
+ rady *= 0.7;
+ TString name = Form("activation%f%f", cx, cy);
+ TPad* p = new TPad(name+"", name+"", cx-radx, cy-rady, cx+radx, cy+rady);
+
+ p->Draw();
+ p->cd();
+
+ activation->Draw();
+ c->cd();
+}
+
+void draw_layer(TCanvas* c, TH2F* h, Int_t iHist,
+ Int_t nLayers, Double_t maxWeight)
+{
+ const Double_t MAX_NEURONS_NICE = 12;
+ const Double_t LABEL_HEIGHT = 0.03;
+ const Double_t LABEL_WIDTH = 0.20;
+ Double_t ratio = ((Double_t)(c->GetWindowHeight())) / c->GetWindowWidth();
+ Double_t rad, cx1, *cy1, cx2, *cy2;
+
+ // this is the smallest radius that will still display the activation images
+ rad = 0.04*650/c->GetWindowHeight();
+
+ Int_t nNeurons1 = h->GetNbinsX();
+ cx1 = iHist*(1.0-LABEL_WIDTH)/nLayers + 1.0/(2.0*nLayers) + LABEL_WIDTH;
+ cy1 = new Double_t[nNeurons1];
+
+ Int_t nNeurons2 = h->GetNbinsY();
+ cx2 = (iHist+1)*(1.0-LABEL_WIDTH)/nLayers + 1.0/(2.0*nLayers) + LABEL_WIDTH;
+ cy2 = new Double_t[nNeurons2];
+
+ Double_t effRad1 = rad;
+ if (nNeurons1 > MAX_NEURONS_NICE)
+ effRad1 = 0.8*(1.0-LABEL_HEIGHT)/(2.0*nNeurons1);
+
+
+ for (Int_t i = 0; i < nNeurons1; i++) {
+ cy1[nNeurons1-i-1] = i*(1.0-LABEL_HEIGHT)/nNeurons1 +
+ 1.0/(2.0*nNeurons1) + LABEL_HEIGHT;
+
+ if (iHist == 0) {
+
+ TEllipse *ellipse
+ = new TEllipse(cx1, cy1[nNeurons1-i-1],
+ effRad1*ratio, effRad1, 0, 360, 0);
+ ellipse->Draw();
+
+ if (i == 0) ellipse->SetLineColor(9);
+
+ if (nNeurons1 > MAX_NEURONS_NICE) continue;
+
+ Int_t whichActivation = 0;
+ if (iHist==0 || iHist==nLayers-1 || i==0) whichActivation = 1;
+ draw_activation(c, cx1, cy1[nNeurons1-i-1],
+ rad*ratio, rad, whichActivation);
+ }
+ }
+
+ if (iHist == 0) draw_input_labels(nNeurons1, cy1, rad, (1.0-LABEL_WIDTH)/nLayers);
+
+ Double_t effRad2 = rad;
+ if (nNeurons2 > MAX_NEURONS_NICE)
+ effRad2 = 0.8*(1.0-LABEL_HEIGHT)/(2.0*nNeurons2);
+
+ for (Int_t i = 0; i < nNeurons2; i++) {
+ cy2[nNeurons2-i-1] = i*(1.0-LABEL_HEIGHT)/nNeurons2 + 1.0/(2.0*nNeurons2) + LABEL_HEIGHT;
+
+ TEllipse *ellipse =
+ new TEllipse(cx2, cy2[nNeurons2-i-1], effRad2*ratio, effRad2, 0, 360, 0);
+ ellipse->Draw();
+
+ if (i == 0 && nNeurons2 > 1) ellipse->SetLineColor(9);
+
+ if (nNeurons2 > MAX_NEURONS_NICE) continue;
+
+ Int_t whichActivation = 0;
+ if (iHist+1==0 || iHist+1==nLayers-1 || i==0) whichActivation = 1;
+ draw_activation(c, cx2, cy2[nNeurons2-i-1], rad*ratio, rad, whichActivation);
+ }
+
+ for (Int_t i = 0; i < nNeurons1; i++) {
+ for (Int_t j = 0; j < nNeurons2; j++) {
+ draw_synapse(cx1, cy1[i], cx2, cy2[j], effRad1*ratio, effRad2*ratio,
+ h->GetBinContent(i+1, j+1)/maxWeight);
+ }
+ }
+
+ delete[] cy1;
+ delete[] cy2;
+}
+
+void draw_synapse(Double_t cx1, Double_t cy1, Double_t cx2, Double_t cy2,
+ Double_t rad1, Double_t rad2, Double_t weightNormed)
+{
+ const Double_t TIP_SIZE = 0.01;
+ const Double_t MAX_WEIGHT = 8;
+ const Double_t MAX_COLOR = 100; // red
+ const Double_t MIN_COLOR = 60; // blue
+
+ if (weightNormed == 0) return;
+
+ gStyle->SetPalette(100, NULL);
+
+ TArrow *arrow = new TArrow(cx1+rad1, cy1, cx2-rad2, cy2, TIP_SIZE, ">");
+ arrow->SetFillColor(1);
+ arrow->SetFillStyle(1001);
+ arrow->SetLineWidth((Int_t)(TMath::Abs(weightNormed)*MAX_WEIGHT+0.5));
+ arrow->SetLineColor((Int_t)((weightNormed+1.0)/2.0*(MAX_COLOR-MIN_COLOR)+MIN_COLOR+0.5));
+ arrow->Draw();
+}
diff --git a/tmva/test/plotall.C b/tmva/test/plotall.C
new file mode 100644
index 00000000000..582effcf763
--- /dev/null
+++ b/tmva/test/plotall.C
@@ -0,0 +1,23 @@
+#include "variables.C"
+#include "correlations.C"
+#include "efficiencies.C"
+#include "mvas.C"
+#include "mutransform.C"
+
+void plotall( TString fin = "TMVA.root" )
+{
+ cout << "=== execute: variables()" << endl;
+ variables( fin );
+
+ cout << "=== execute: correlations()" << endl;
+ correlations( fin );
+
+ cout << "=== execute: mvas()" << endl;
+ mvas( fin );
+
+ cout << "=== execute: efficiencies()" << endl;
+ efficiencies( fin );
+
+ cout << "=== execute: ztransform()" << endl;
+ mutransform( fin );
+}
diff --git a/tmva/test/sigmoid-small.png b/tmva/test/sigmoid-small.png
new file mode 100644
index 0000000000000000000000000000000000000000..c5882fb8942f829d03540953c87b3678c3ce3d2b
GIT binary patch
literal 363
zcmeAS@N?(olHy`uVBq!ia0y~yU@!t<4kiW$hKaHA&lngO7>k44ofy`glX=O&z`&C3
z=<CS9u(6-}Pa-P=1A{`cN02WALzNl>LqiJ#!!HH~hK3gm45bDP46hOx7_4S6Fo+k-
z*%fHRz`*d*)5S3)!u{=yhg{7DJS-2s{*~YLip%xp1n=sdBEE_LU;U{PFIjUX%KqZ&
zDMe~BOa2OPwR$QpaA4QPlKHc?&Yn=T!1si>VztBMl5GW(JFd2Ib_SbAhXqW&$(LwV
zSgjPLU#XpKp><p+-A6y!*X`-lYh7z9CwoU}Z_-e!i#pkKKWoj36RK+3iyYHBkFaIx
ztkaDA&~vHI<g)aWDKm5`L~Ir*O_{Og1zWmzNmfDE({!oyowMd0{nVFbuvRpZX|knm
z@5y76X7cDP<xK7L74XjAm0bJ(k-gZTM`?L`?raTSWb!v9-P6aQ-uWIwY{K&y+=BVG
Q3=9kmp00i_>zopr0GUvV`2YX_
literal 0
HcmV?d00001
diff --git a/tmva/test/tmvaglob.C b/tmva/test/tmvaglob.C
index 0de08fe7721..58e020648ac 100644
--- a/tmva/test/tmvaglob.C
+++ b/tmva/test/tmvaglob.C
@@ -1,114 +1,169 @@
// global TMVA style settings
+#ifndef TMVA_TMVAGLOB
+#define TMVA_TMVAGLOB
namespace TMVAGlob {
- //signal
- const Int_t FillColor__S = 38;
- const Int_t FillStyle__S = 1001;
- const Int_t LineColor__S = 1;
- const Int_t LineWidth__S = 1;
-
- // background
- const Int_t FillColor__B = 2;
- const Int_t FillStyle__B = 3002;
- const Int_t LineColor__B = 2;
- const Int_t LineWidth__B = 1;
-
- // set the style
- void SetSignalAndBackgroundStyle( TH1* sig, TH1* bgd, TH1* all = 0 )
- {
- if (sig != NULL) {
- sig->SetLineColor( LineColor__S );
- sig->SetLineWidth( LineWidth__S );
- sig->SetFillStyle( FillStyle__S );
- sig->SetFillColor( FillColor__S );
- }
+ enum TypeOfPlot { kNormal = 0,
+ kDecorrelated,
+ kPCA,
+ kNumOfMethods };
+
+
+ // set the style
+ void SetSignalAndBackgroundStyle( TH1* sig, TH1* bgd, TH1* all = 0 )
+ {
+
+ //signal
+ const Int_t FillColor__S = 38;
+ const Int_t FillStyle__S = 1001;
+ const Int_t LineColor__S = 1;
+ const Int_t LineWidth__S = 2;
+
+ // background
+ const Int_t FillColor__B = 2;
+ const Int_t FillStyle__B = 3554;
+ const Int_t LineColor__B = 2;
+ const Int_t LineWidth__B = 2;
+
+ if (sig != NULL) {
+ sig->SetLineColor( LineColor__S );
+ sig->SetLineWidth( LineWidth__S );
+ sig->SetFillStyle( FillStyle__S );
+ sig->SetFillColor( FillColor__S );
+ }
- if (bgd != NULL) {
- bgd->SetLineColor( LineColor__B );
- bgd->SetLineWidth( LineWidth__B );
- bgd->SetFillStyle( FillStyle__B );
- bgd->SetFillColor( FillColor__B );
- }
-
- if (all != NULL) {
- all->SetLineColor( LineColor__S );
- all->SetLineWidth( LineWidth__S );
- all->SetFillStyle( FillStyle__S );
- all->SetFillColor( FillColor__S );
- }
- }
-
- // set frame styles
- SetFrameStyle( TH1* frame, Float_t scale = 1.0 )
- {
- frame->SetLabelOffset( 0.012, "X" );// label offset on x axis
- frame->SetLabelOffset( 0.012, "Y" );// label offset on x axis
- frame->GetXaxis()->SetTitleOffset( 1.25 );
- frame->GetYaxis()->SetTitleOffset( 1.22 );
- frame->GetXaxis()->SetTitleSize( 0.045*scale );
- frame->GetYaxis()->SetTitleSize( 0.045*scale );
- Float_t labelSize = 0.04*scale;
- frame->GetXaxis()->SetLabelSize( labelSize );
- frame->GetYaxis()->SetLabelSize( labelSize );
-
- // global style settings
- gPad->SetTicks();
- gPad->SetLeftMargin ( 0.108*scale );
- gPad->SetRightMargin ( 0.050*scale );
- gPad->SetBottomMargin( 0.120*scale );
- }
-
- // used to create output file for canvas
- void imgconv( TCanvas* c, TString fname )
- {
- if (NULL == c) {
- cout << "--- Error in TMVAGlob::imgconv: canvas is NULL" << endl;
- }
- else {
- // create directory if not existing
- TString f = fname;
- TString dir = f.Remove( f.Last( '/' ), f.Length() - f.Last( '/' ) );
- gSystem->mkdir( dir );
-
- TString pngName = fname + ".png";
- TString gifName = fname + ".gif";
- TString epsName = fname + ".eps";
-
- // create png
- TImage *img = TImage::Create();
- img->FromPad( c );
- img->WriteImage( gifName );
- img->WriteImage( pngName );
-
- // create eps (other option: c->Print( epsName ))
- c->SaveAs(epsName);
-
- cout << "--- Create output files: " << pngName << " (+ .gif and .eps)" << endl;
- }
- }
-
- void plot_logo( Float_t v_scale = 1.0 )
- {
- TImage *img = TImage::Open("tmva_logo.gif");
- if (!img) {
- printf("Could not create an image... exit\n");
- return;
- }
- img->SetConstRatio(kFALSE);
- UInt_t h_ = img->GetHeight();
- UInt_t w_ = img->GetWidth();
-
- Float_t d = 0.045;
- // absolute coordinates
- Float_t x1L = 0.7803;
- Float_t y1L = 0.91;
- TPad *p1 = new TPad("img", "img", x1L, y1L, x1L + d*w_/h_, y1L + d*1.5*v_scale );
-
- p1->Draw();
- p1->cd();
- img->Draw();
- }
+ if (bgd != NULL) {
+ bgd->SetLineColor( LineColor__B );
+ bgd->SetLineWidth( LineWidth__B );
+ bgd->SetFillStyle( FillStyle__B );
+ bgd->SetFillColor( FillColor__B );
+ }
+
+ if (all != NULL) {
+ all->SetLineColor( LineColor__S );
+ all->SetLineWidth( LineWidth__S );
+ all->SetFillStyle( FillStyle__S );
+ all->SetFillColor( FillColor__S );
+ }
+ }
+
+ // set frame styles
+ SetFrameStyle( TH1* frame, Float_t scale = 1.0 )
+ {
+ frame->SetLabelOffset( 0.012, "X" );// label offset on x axis
+ frame->SetLabelOffset( 0.012, "Y" );// label offset on x axis
+ frame->GetXaxis()->SetTitleOffset( 1.25 );
+ frame->GetYaxis()->SetTitleOffset( 1.22 );
+ frame->GetXaxis()->SetTitleSize( 0.045*scale );
+ frame->GetYaxis()->SetTitleSize( 0.045*scale );
+ Float_t labelSize = 0.04*scale;
+ frame->GetXaxis()->SetLabelSize( labelSize );
+ frame->GetYaxis()->SetLabelSize( labelSize );
+
+ // global style settings
+ gPad->SetTicks();
+ gPad->SetLeftMargin ( 0.108*scale );
+ gPad->SetRightMargin ( 0.050*scale );
+ gPad->SetBottomMargin( 0.120*scale );
+ }
+
+ // set style and remove existing canvas'
+ void Initialize( Bool_t useTMVAStyle = kTRUE )
+ {
+ // set style
+ if (!useTMVAStyle) {
+ gROOT->SetStyle("Plain");
+ gStyle->SetOptStat(0);
+ }
+
+ // destroy canvas'
+ TList * loc = gROOT->GetListOfCanvases();
+ TListIter itc(loc);
+ TObject *o(0);
+ while ((o = itc())) delete o;
+ }
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* OpenFile( const TString& fin )
+ {
+ TFile* file = gDirectory->GetFile();
+ if (file==0 || fin != file->GetName()) {
+ if (file != 0) {
+ gROOT->cd();
+ file->Close();
+ }
+ cout << "Creating TFile: " << fin << endl;
+ file = new TFile( fin );
+ }
+ else {
+ cout << "Retrieving TFile: " << fin << endl;
+ file = gDirectory->GetFile();
+ }
+
+ file->cd();
+ return file;
+ }
+
+ // used to create output file for canvas
+ void imgconv( TCanvas* c, TString fname )
+ {
+ // return;
+ if (NULL == c) {
+ cout << "--- Error in TMVAGlob::imgconv: canvas is NULL" << endl;
+ }
+ else {
+ // create directory if not existing
+ TString f = fname;
+ TString dir = f.Remove( f.Last( '/' ), f.Length() - f.Last( '/' ) );
+ gSystem->mkdir( dir );
+
+ TString pngName = fname + ".png";
+ TString gifName = fname + ".gif";
+ TString epsName = fname + ".eps";
+
+ // create eps (other option: c->Print( epsName ))
+ c->SaveAs(epsName);
+ cout << "If you want to save the image as gif or png, please comment out "
+ << "the corresponding lines (line no. 83+84) in tmvaglob.C" << endl;
+ // c->SaveAs(gifName);
+ c->SaveAs(pngName);
+ }
+ }
+
+ void plot_logo( Float_t v_scale = 1.0 )
+ {
+ TImage *img = TImage::Open("tmva_logo.gif");
+ if (!img) {
+ printf("Could not create an image... exit\n");
+ return;
+ }
+ img->SetConstRatio(kFALSE);
+ UInt_t h_ = img->GetHeight();
+ UInt_t w_ = img->GetWidth();
+ //cout << w_/h_ << endl;
+
+ Float_t rgif = 405/108.;
+ Float_t rpad = gPad->GetWw()/gPad->GetWh();
+ Float_t xperc = 0.3;
+ Float_t yperc = xperc * rpad / rgif;
+
+ Float_t r = w_/h_;
+
+ Float_t d = 0.045;
+ // absolute coordinates
+ Float_t x1L = 1 - gStyle->GetPadRightMargin();
+ Float_t y1L = 0.91;
+ TPad *p1 = new TPad("img", "img", x1L - d*r, y1L, x1L, y1L + d*1.5*v_scale );
+ p1->SetRightMargin(0);
+ p1->SetBottomMargin(0);
+ p1->SetLeftMargin(0);
+ p1->SetTopMargin(0);
+ p1->Draw();
+ p1->cd();
+ img->Draw();
+ }
}
+#endif
diff --git a/tmva/test/variables.C b/tmva/test/variables.C
index 26a2b6a4dda..1fdcd09cc19 100644
--- a/tmva/test/variables.C
+++ b/tmva/test/variables.C
@@ -1,144 +1,166 @@
#include "tmvaglob.C"
-void variables( TString fin = "TMVA.root" )
+// this macro plots the distributions of the different input variables
+// used in TMVA (e.g. running TMVAnalysis.C). Signal and Background are overlayed.
+
+
+
+const TString directories[TMVAGlob::kNumOfMethods] = { "input_variables",
+ "decorrelated_input_variables",
+ "principal_component_analyzed_input_variables" };
+
+const TString titles[TMVAGlob::kNumOfMethods] = { "TMVA Input Variables",
+ "Decorrelated TMVA Input Variables",
+ "Principal Component Transformed TMVA Input Variables" };
+
+const TString outfname[TMVAGlob::kNumOfMethods] = { "variables",
+ "variables_decorr",
+ "variables_pca" };
+
+// input: - Input file (result from TMVA),
+// - normal/decorrelated/PCA
+// - use of TMVA plotting TStyle
+void variables( TString fin = "TMVA.root", TMVAGlob::TypeOfPlot type = TMVAGlob::kNormal, bool useTMVAStyle=kTRUE )
{
- gROOT->Reset();
- gROOT->SetStyle("Plain");
- gStyle->SetOptStat(0);
- TList * loc = gROOT->GetListOfCanvases();
- TListIter itc(loc);
- TObject *o(0);
- while( (o = itc()) ) delete o;
-
- TFile *file = new TFile( fin );
-
- input_variables->cd();
-
- // how many plots are in the "input_variables" directory?
- Int_t noPlots = ((gDirectory->GetListOfKeys())->GetEntries()) / 2;
-
- // define Canvas layout here!
- // default setting
- Int_t xPad; // no of plots in x
- Int_t yPad; // no of plots in y
- Int_t width; // size of canvas
- Int_t height;
- switch (noPlots) {
- case 1:
- xPad = 1; yPad = 1; width = 500; height = width; break;
- case 2:
- xPad = 2; yPad = 1; width = 600; height = 0.7*width; break;
- case 3:
- xPad = 3; yPad = 1; width = 800; height = 0.5*width; break;
- case 4:
- xPad = 2; yPad = 2; width = 600; height = width; break;
- default:
- xPad = 3; yPad = 2; width = 800; height = 0.7*width; break;
- }
- Int_t noPad = xPad * yPad ;
-
- // this defines how many canvases we need
- const Int_t noCanvas = 1 + (Int_t)(noPlots/noPad);
- TCanvas **c = new TCanvas*[noCanvas];
- for (Int_t ic=0; ic<noCanvas; ic++) c[ic] = 0;
-
- cout << "--- Found: " << noPlots << " plots; "
- << "will produce: " << noCanvas << " canvas" << endl;
-
- // counter variables
- Int_t countCanvas = 0;
- Int_t countPad = 1;
-
- // loop over all objects in "input_variables" directory
- TIter next(gDirectory->GetListOfKeys());
- TKey *key;
- char fname[200];
- while ((key = (TKey*)next())) {
-
- // make sure, that we only look at histograms
- TClass *cl = gROOT->GetClass(key->GetClassName());
- if (!cl->InheritsFrom("TH1")) continue;
- TH1 *sig = (TH1*)key->ReadObj();
- TString hname= sig->GetName();
-
- // check for all signal histograms
- if (hname.Contains("__S")){ // found a new signal plot
-
- // create new canvas
- if ((c[countCanvas]==NULL) || (countPad>noPad)) {
- cout << "--- Book canvas no: " << countCanvas << endl;
- char cn[20];
- sprintf( cn, "canvas%d", countCanvas+1 );
- c[countCanvas] = new TCanvas( cn, "MVA Input Variables",
- countCanvas*50+300, countCanvas*20, width, height );
- // style
- c[countCanvas]->SetBorderMode(0);
- c[countCanvas]->SetFillColor(10);
-
- c[countCanvas]->Divide(xPad,yPad);
- countPad = 1;
- }
-
- // save canvas to file
- c[countCanvas]->cd(countPad);
- countPad++;
- if (countPad > noPad) {
- c[countCanvas]->Update();
- sprintf( fname, "plots/variables_c%i", countCanvas+1 );
- TMVAGlob::imgconv( c[countCanvas], &fname[0] );
- countCanvas++;
+ // set style and remove existing canvas'
+ TMVAGlob::Initialize( useTMVAStyle );
+
+ // checks if file with name "fin" is already open, and if not opens one
+ TFile* file = TMVAGlob::OpenFile( fin );
+
+ TDirectory* dir = (TDirectory*)gDirectory->Get( directories[type] );
+ if (dir==0) {
+ cout << "Could not locate directory '" << directories[type] << "' in file: " << fin << endl;
+ return;
+ }
+ dir->cd();
+
+ // how many plots are in the directory?
+ Int_t noPlots = ((dir->GetListOfKeys())->GetEntries()) / 2;
+
+ // define Canvas layout here!
+ // default setting
+ Int_t xPad; // no of plots in x
+ Int_t yPad; // no of plots in y
+ Int_t width; // size of canvas
+ Int_t height;
+ switch (noPlots) {
+ case 1:
+ xPad = 1; yPad = 1; width = 500; height = width; break;
+ case 2:
+ xPad = 2; yPad = 1; width = 600; height = 0.7*width; break;
+ case 3:
+ xPad = 3; yPad = 1; width = 800; height = 0.5*width; break;
+ case 4:
+ xPad = 2; yPad = 2; width = 400; height = width; break;
+ default:
+ xPad = 3; yPad = 2; width = 800; height = 0.7*width; break;
+ }
+ Int_t noPad = xPad * yPad ;
+
+ // this defines how many canvases we need
+ const Int_t noCanvas = 1 + (Int_t)((noPlots - 0.001)/noPad);
+ TCanvas **c = new TCanvas*[noCanvas];
+ for (Int_t ic=0; ic<noCanvas; ic++) c[ic] = 0;
+
+ cout << "--- Found: " << noPlots << " plots; will produce: " << noCanvas << " canva(s)" << endl;
+
+ // counter variables
+ Int_t countCanvas = 0;
+ Int_t countPad = 1;
+
+ // loop over all objects in directory
+ TIter next(dir->GetListOfKeys());
+ TKey *key;
+ while ((key = (TKey*)next())) {
+
+ // make sure, that we only look at histograms
+ TClass *cl = gROOT->GetClass(key->GetClassName());
+ if (!cl->InheritsFrom("TH1")) continue;
+ TH1 *sig = (TH1*)key->ReadObj();
+ TString hname= sig->GetName();
+
+ // check for all signal histograms
+ if (hname.Contains("__S")){ // found a new signal plot
+
+ // create new canvas
+ if ((c[countCanvas]==NULL) || (countPad>noPad)) {
+ cout << "--- Book canvas no: " << countCanvas << endl;
+ char cn[20];
+ sprintf( cn, "canvas%d", countCanvas+1 );
+ c[countCanvas] = new TCanvas( cn, titles[type],
+ countCanvas*50+200, countCanvas*20, width, height );
+ // style
+ c[countCanvas]->SetBorderMode(0);
+ c[countCanvas]->SetFillColor(0);
+
+ c[countCanvas]->Divide(xPad,yPad);
+ countPad = 1;
+ }
+
+ // save canvas to file
+ c[countCanvas]->cd(countPad);
+ countPad++;
+
+ // find the corredponding backgrouns histo
+ TString bgname = hname;
+ bgname.ReplaceAll("__S","__B");
+ TH1 *bgd = (TH1*)dir->Get(bgname);
+ if (bgd == NULL) {
+ cout << "ERROR!!! couldn't find backgroung histo for" << hname << endl;
+ exit;
+ }
+
+ // this is set but not stored during plot creation in MVA_Factory
+ TMVAGlob::SetSignalAndBackgroundStyle( sig, bgd );
+
+ // chop off "signal"
+ TString title(sig->GetTitle());
+ title.ReplaceAll("signal","");
+ sig->SetTitle( TString( titles[type] ) + title );
+ TMVAGlob::SetFrameStyle( sig, 1.2 );
+
+ // finally plot and overlay
+ Float_t sc = 1.1;
+ if (countPad==2) sc = 1.3;
+ sig->SetMaximum( TMath::Max( sig->GetMaximum(), bgd->GetMaximum() )*sc );
+ sig->Draw();
+
+ bgd->Draw("same");
+ sig->GetXaxis()->SetTitle( title );
+ sig->GetYaxis()->SetTitleOffset( 1.35 );
+ sig->GetYaxis()->SetTitle("Normalized");
+
+ // redraw axes
+ sig->Draw("sameaxis");
+
+ // Draw legend
+ if (countPad==2){
+ TLegend *legend= new TLegend( gPad->GetLeftMargin(),
+ 1-gPad->GetTopMargin()-.18,
+ gPad->GetLeftMargin()+.4,
+ 1-gPad->GetTopMargin() );
+ legend->AddEntry(sig,"Signal","F");
+ legend->AddEntry(bgd,"Background","F");
+ legend->Draw("same");
+ legend->SetBorderSize(1);
+ legend->SetMargin( 0.3 );
+ }
+
+ // save canvas to file
+ if (countPad > noPad) {
+ c[countCanvas]->Update();
+ TString fname = Form( "plots/%s_c%i", outfname[type].Data(), countCanvas+1 );
+ TMVAGlob::imgconv( c[countCanvas], &fname[0] );
+ // TMVAGlob::plot_logo(); // don't understand why this doesn't work ... :-(
+ countCanvas++;
+ }
}
+ }
- // find the corredponding backgrouns histo
- TString bgname = hname;
- bgname.ReplaceAll("__S","__B");
- TH1 *bgd = (TH1*)gDirectory->Get(bgname);
- if (bgd == NULL){
- cout << "ERROR!!! couldn't find backgroung histo for" << hname << endl;
- exit;
- }
- // this is set but not stored during plot creation in MVA_Factory
- TMVAGlob::SetSignalAndBackgroundStyle( sig, bgd );
-
- // chop off "signal"
- TString title(sig->GetTitle());
- title.ReplaceAll("signal","");
- sig->SetTitle( TString("MVA input variable: ") + title );
-
- TMVAGlob::SetFrameStyle( sig, 1.2 );
-
- // finally plot and overlay
- Float_t sc = 1.1;
- if (countPad==2) sc = 1.3;
- sig->SetMaximum( TMath::Max( sig->GetMaximum(), bgd->GetMaximum() )*sc );
- sig->Draw();
- bgd->SetLineColor( 2 );
- bgd->SetLineWidth( 1 );
- bgd->SetFillStyle( 3002 );
- bgd->SetFillColor( 46 );
- bgd->Draw("same");
- sig->GetYaxis()->SetTitleOffset( 1.35 );
- sig->GetYaxis()->SetTitle("Normalized");
-
- // redraw axes
- sig->Draw("sameaxis");
-
- // Draw legend
- if (countPad==2){
- TLegend *legend= new TLegend( 0.131, 0.762, 0.531, 0.901 );
- legend->AddEntry(sig,"Signal","F");
- legend->AddEntry(bgd,"Background","F");
- legend->Draw("same");
- legend->SetBorderSize(1);
- legend->SetMargin( 0.3 );
- }
-
- }
- }
- if (countPad <= noPad) {
-
- c[countCanvas]->Update();
- sprintf( fname, "plots/variables_c%i", countCanvas+1 );
- TMVAGlob::imgconv( c[countCanvas], &fname[0] );
- }
+ if (countPad <= noPad) {
+ c[countCanvas]->Update();
+ TString fname = Form( "plots/%s_c%i", outfname[type].Data(), countCanvas+1 );
+ TMVAGlob::imgconv( c[countCanvas], &fname[0] );
+ }
}
--
GitLab