/usr/include/root/TMVA/MethodRuleFit.h is in libroot-tmva-dev 5.34.30-0ubuntu8.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 | // @(#)root/tmva $Id$
// Author: Fredrik Tegenfeldt
/**********************************************************************************
* Project: TMVA - a Root-integrated toolkit for multivariate data analysis *
* Package: TMVA *
* Class : MethodRuleFit *
* Web : http://tmva.sourceforge.net *
* *
* Description: *
* Friedman's RuleFit method *
* *
* Authors (alphabetical): *
* Fredrik Tegenfeldt <Fredrik.Tegenfeldt@cern.ch> - Iowa State U., USA *
* *
* Copyright (c) 2005: *
* CERN, Switzerland *
* Iowa State U. *
* 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 *
* *
**********************************************************************************/
#ifndef ROOT_TMVA_MethodRuleFit
#define ROOT_TMVA_MethodRuleFit
//////////////////////////////////////////////////////////////////////////
// //
// MethodRuleFit //
// //
// J Friedman's RuleFit method //
// //
//////////////////////////////////////////////////////////////////////////
#ifndef ROOT_TMVA_MethodBase
#include "TMVA/MethodBase.h"
#endif
#ifndef ROOT_TMatrixDfwd
#include "TMatrixDfwd.h"
#endif
#ifndef ROOT_TVectorD
#include "TVectorD.h"
#endif
#ifndef ROOT_TMVA_DecisionTree
#include "TMVA/DecisionTree.h"
#endif
#ifndef ROOT_TMVA_RuleFit
#include "TMVA/RuleFit.h"
#endif
namespace TMVA {
class SeparationBase;
class MethodRuleFit : public MethodBase {
public:
MethodRuleFit( const TString& jobName,
const TString& methodTitle,
DataSetInfo& theData,
const TString& theOption = "",
TDirectory* theTargetDir = 0 );
MethodRuleFit( DataSetInfo& theData,
const TString& theWeightFile,
TDirectory* theTargetDir = NULL );
virtual ~MethodRuleFit( void );
virtual Bool_t HasAnalysisType( Types::EAnalysisType type, UInt_t numberClasses, UInt_t /*numberTargets*/ );
// training method
void Train( void );
using MethodBase::ReadWeightsFromStream;
// write weights to file
void AddWeightsXMLTo ( void* parent ) const;
// read weights from file
void ReadWeightsFromStream( std::istream& istr );
void ReadWeightsFromXML ( void* wghtnode );
// calculate the MVA value
Double_t GetMvaValue( Double_t* err = 0, Double_t* errUpper = 0 );
// write method specific histos to target file
void WriteMonitoringHistosToFile( void ) const;
// ranking of input variables
const Ranking* CreateRanking();
Bool_t UseBoost() const { return fUseBoost; }
// accessors
RuleFit* GetRuleFitPtr() { return &fRuleFit; }
const RuleFit* GetRuleFitConstPtr() const { return &fRuleFit; }
TDirectory* GetMethodBaseDir() const { return BaseDir(); }
const std::vector<TMVA::Event*>& GetTrainingEvents() const { return fEventSample; }
const std::vector<TMVA::DecisionTree*>& GetForest() const { return fForest; }
Int_t GetNTrees() const { return fNTrees; }
Double_t GetTreeEveFrac() const { return fTreeEveFrac; }
const SeparationBase* GetSeparationBaseConst() const { return fSepType; }
SeparationBase* GetSeparationBase() const { return fSepType; }
TMVA::DecisionTree::EPruneMethod GetPruneMethod() const { return fPruneMethod; }
Double_t GetPruneStrength() const { return fPruneStrength; }
Double_t GetMinFracNEve() const { return fMinFracNEve; }
Double_t GetMaxFracNEve() const { return fMaxFracNEve; }
Int_t GetNCuts() const { return fNCuts; }
//
Int_t GetGDNPathSteps() const { return fGDNPathSteps; }
Double_t GetGDPathStep() const { return fGDPathStep; }
Double_t GetGDErrScale() const { return fGDErrScale; }
Double_t GetGDPathEveFrac() const { return fGDPathEveFrac; }
Double_t GetGDValidEveFrac() const { return fGDValidEveFrac; }
//
Double_t GetLinQuantile() const { return fLinQuantile; }
const TString GetRFWorkDir() const { return fRFWorkDir; }
Int_t GetRFNrules() const { return fRFNrules; }
Int_t GetRFNendnodes() const { return fRFNendnodes; }
protected:
// make ROOT-independent C++ class for classifier response (classifier-specific implementation)
void MakeClassSpecific( std::ostream&, const TString& ) const;
void MakeClassRuleCuts( std::ostream& ) const;
void MakeClassLinear( std::ostream& ) const;
// get help message text
void GetHelpMessage() const;
// initialize rulefit
void Init( void );
// copy all training events into a stl::vector
void InitEventSample( void );
// initialize monitor ntuple
void InitMonitorNtuple();
void TrainTMVARuleFit();
void TrainJFRuleFit();
private:
// check variable range and set var to lower or upper if out of range
template<typename T>
inline Bool_t VerifyRange( MsgLogger& mlog, const char *varstr, T& var, const T& vmin, const T& vmax );
template<typename T>
inline Bool_t VerifyRange( MsgLogger& mlog, const char *varstr, T& var, const T& vmin, const T& vmax, const T& vdef );
template<typename T>
inline Int_t VerifyRange( const T& var, const T& vmin, const T& vmax );
// the option handling methods
void DeclareOptions();
void ProcessOptions();
RuleFit fRuleFit; // RuleFit instance
std::vector<TMVA::Event *> fEventSample; // the complete training sample
Double_t fSignalFraction; // scalefactor for bkg events to modify initial s/b fraction in training data
// ntuple
TTree *fMonitorNtuple; // pointer to monitor rule ntuple
Double_t fNTImportance; // ntuple: rule importance
Double_t fNTCoefficient; // ntuple: rule coefficient
Double_t fNTSupport; // ntuple: rule support
Int_t fNTNcuts; // ntuple: rule number of cuts
Int_t fNTNvars; // ntuple: rule number of vars
Double_t fNTPtag; // ntuple: rule P(tag)
Double_t fNTPss; // ntuple: rule P(tag s, true s)
Double_t fNTPsb; // ntuple: rule P(tag s, true b)
Double_t fNTPbs; // ntuple: rule P(tag b, true s)
Double_t fNTPbb; // ntuple: rule P(tag b, true b)
Double_t fNTSSB; // ntuple: rule S/(S+B)
Int_t fNTType; // ntuple: rule type (+1->signal, -1->bkg)
// options
TString fRuleFitModuleS;// which rulefit module to use
Bool_t fUseRuleFitJF; // if true interface with J.Friedmans RuleFit module
TString fRFWorkDir; // working directory from Friedmans module
Int_t fRFNrules; // max number of rules (only Friedmans module)
Int_t fRFNendnodes; // max number of rules (only Friedmans module)
std::vector<DecisionTree *> fForest; // the forest
Int_t fNTrees; // number of trees in forest
Double_t fTreeEveFrac; // fraction of events used for traing each tree
SeparationBase *fSepType; // the separation used in node splitting
Double_t fMinFracNEve; // min fraction of number events
Double_t fMaxFracNEve; // ditto max
Int_t fNCuts; // grid used in cut applied in node splitting
TString fSepTypeS; // forest generation: separation type - see DecisionTree
TString fPruneMethodS; // forest generation: prune method - see DecisionTree
TMVA::DecisionTree::EPruneMethod fPruneMethod; // forest generation: method used for pruning - see DecisionTree
Double_t fPruneStrength; // forest generation: prune strength - see DecisionTree
TString fForestTypeS; // forest generation: how the trees are generated
Bool_t fUseBoost; // use boosted events for forest generation
//
Double_t fGDPathEveFrac; // GD path: fraction of subsamples used for the fitting
Double_t fGDValidEveFrac; // GD path: fraction of subsamples used for the fitting
Double_t fGDTau; // GD path: def threshhold fraction [0..1]
Double_t fGDTauPrec; // GD path: precision of estimated tau
Double_t fGDTauMin; // GD path: min threshhold fraction [0..1]
Double_t fGDTauMax; // GD path: max threshhold fraction [0..1]
UInt_t fGDTauScan; // GD path: number of points to scan
Double_t fGDPathStep; // GD path: step size in path
Int_t fGDNPathSteps; // GD path: number of steps
Double_t fGDErrScale; // GD path: stop
Double_t fMinimp; // rule/linear: minimum importance
//
TString fModelTypeS; // rule ensemble: which model (rule,linear or both)
Double_t fRuleMinDist; // rule min distance - see RuleEnsemble
Double_t fLinQuantile; // quantile cut to remove outliers - see RuleEnsemble
ClassDef(MethodRuleFit,0) // Friedman's RuleFit method
};
} // namespace TMVA
//_______________________________________________________________________
template<typename T>
inline Int_t TMVA::MethodRuleFit::VerifyRange( const T& var, const T& vmin, const T& vmax )
{
// check range and return +1 if above, -1 if below or 0 if inside
if (var>vmax) return 1;
if (var<vmin) return -1;
return 0;
}
//_______________________________________________________________________
template<typename T>
inline Bool_t TMVA::MethodRuleFit::VerifyRange( TMVA::MsgLogger& mlog, const char *varstr, T& var, const T& vmin, const T& vmax )
{
// verify range and print out message
// if outside range, set to closest limit
Int_t dir = TMVA::MethodRuleFit::VerifyRange(var,vmin,vmax);
Bool_t modif=kFALSE;
if (dir==1) {
modif = kTRUE;
var=vmax;
}
if (dir==-1) {
modif = kTRUE;
var=vmin;
}
if (modif) {
mlog << kWARNING << "Option <" << varstr << "> " << (dir==1 ? "above":"below") << " allowed range. Reset to new value = " << var << Endl;
}
return modif;
}
//_______________________________________________________________________
template<typename T>
inline Bool_t TMVA::MethodRuleFit::VerifyRange( TMVA::MsgLogger& mlog, const char *varstr, T& var, const T& vmin, const T& vmax, const T& vdef )
{
// verify range and print out message
// if outside range, set to given default value
Int_t dir = TMVA::MethodRuleFit::VerifyRange(var,vmin,vmax);
Bool_t modif=kFALSE;
if (dir!=0) {
modif = kTRUE;
var=vdef;
}
if (modif) {
mlog << kWARNING << "Option <" << varstr << "> " << (dir==1 ? "above":"below") << " allowed range. Reset to default value = " << var << Endl;
}
return modif;
}
#endif // MethodRuleFit_H
|