Logo ROOT   6.16/01
Reference Guide
RuleFit.h
Go to the documentation of this file.
1// @(#)root/tmva $Id$
2// Author: Andreas Hoecker, Joerg Stelzer, Fredrik Tegenfeldt, Helge Voss
3
4/**********************************************************************************
5 * Project: TMVA - a Root-integrated toolkit for multivariate data analysis *
6 * Package: TMVA *
7 * Class : RuleFit *
8 * Web : http://tmva.sourceforge.net *
9 * *
10 * Description: *
11 * A class implementing various fits of rule ensembles *
12 * *
13 * Authors (alphabetical): *
14 * Fredrik Tegenfeldt <Fredrik.Tegenfeldt@cern.ch> - Iowa State U., USA *
15 * Helge Voss <Helge.Voss@cern.ch> - MPI-KP Heidelberg, Ger. *
16 * *
17 * Copyright (c) 2005: *
18 * CERN, Switzerland *
19 * Iowa State U. *
20 * MPI-K Heidelberg, Germany *
21 * *
22 * Redistribution and use in source and binary forms, with or without *
23 * modification, are permitted according to the terms listed in LICENSE *
24 * (http://tmva.sourceforge.net/LICENSE) *
25 **********************************************************************************/
26
27#ifndef ROOT_TMVA_RuleFit
28#define ROOT_TMVA_RuleFit
29
30#include "TMVA/DecisionTree.h"
31#include "TMVA/RuleEnsemble.h"
32#include "TMVA/RuleFitParams.h"
33#include "TMVA/Event.h"
34
35#include <algorithm>
36#include <random>
37
38namespace TMVA {
39
40
41 class MethodBase;
42 class MethodRuleFit;
43 class MsgLogger;
44
45 class RuleFit {
46
47 public:
48
49 // main constructor
50 RuleFit( const TMVA::MethodBase *rfbase );
51
52 // empty constructor
53 RuleFit( void );
54
55 virtual ~RuleFit( void );
56
57 void InitNEveEff();
58 void InitPtrs( const TMVA::MethodBase *rfbase );
59 void Initialize( const TMVA::MethodBase *rfbase );
60
61 void SetMsgType( EMsgType t );
62
63 void SetTrainingEvents( const std::vector<const TMVA::Event *> & el );
64
66 {
67 std::shuffle(fTrainingEventsRndm.begin(), fTrainingEventsRndm.end(), fRNGEngine);
68 }
69
70 void SetMethodBase( const MethodBase *rfbase );
71
72 // make the forest of trees for rule generation
73 void MakeForest();
74
75 // build a tree
76 void BuildTree( TMVA::DecisionTree *dt );
77
78 // save event weights
79 void SaveEventWeights();
80
81 // restore saved event weights
83
84 // boost events based on the given tree
85 void Boost( TMVA::DecisionTree *dt );
86
87 // calculate and print some statistics on the given forest
88 void ForestStatistics();
89
90 // calculate the discriminating variable for the given event
91 Double_t EvalEvent( const Event& e );
92
93 // calculate sum of
94 Double_t CalcWeightSum( const std::vector<const TMVA::Event *> *events, UInt_t neve=0 );
95
96 // do the fitting of the coefficients
97 void FitCoefficients();
98
99 // calculate variable and rule importance from a set of events
100 void CalcImportance();
101
102 // set usage of linear term
104 // set usage of rules
106 // set usage of linear term
108 // set minimum importance allowed
110 // set minimum rule distance - see RuleEnsemble
112 // set path related parameters
116 // make visualization histograms
120 void MakeVisHists();
121 void FillVisHistCut(const Rule * rule, std::vector<TH2F *> & hlist);
122 void FillVisHistCorr(const Rule * rule, std::vector<TH2F *> & hlist);
123 void FillCut(TH2F* h2,const TMVA::Rule *rule,Int_t vind);
124 void FillLin(TH2F* h2,Int_t vind);
125 void FillCorr(TH2F* h2,const TMVA::Rule *rule,Int_t v1, Int_t v2);
126 void NormVisHists(std::vector<TH2F *> & hlist);
127 void MakeDebugHists();
128 Bool_t GetCorrVars(TString & title, TString & var1, TString & var2);
129 // accessors
131 Double_t GetNEveEff() const { return fNEveEffTrain; } // reweighted number of events = sum(wi)
132 const Event* GetTrainingEvent(UInt_t i) const { return static_cast< const Event *>(fTrainingEvents[i]); }
133 Double_t GetTrainingEventWeight(UInt_t i) const { return fTrainingEvents[i]->GetWeight(); }
134
135 // const Event* GetTrainingEvent(UInt_t i, UInt_t isub) const { return &(fTrainingEvents[fSubsampleEvents[isub]])[i]; }
136
137 const std::vector< const TMVA::Event * > & GetTrainingEvents() const { return fTrainingEvents; }
138 // const std::vector< Int_t > & GetSubsampleEvents() const { return fSubsampleEvents; }
139
140 // void GetSubsampleEvents(Int_t sub, UInt_t & ibeg, UInt_t & iend) const;
141 void GetRndmSampleEvents(std::vector< const TMVA::Event * > & evevec, UInt_t nevents);
142 //
143 const std::vector< const TMVA::DecisionTree *> & GetForest() const { return fForest; }
144 const RuleEnsemble & GetRuleEnsemble() const { return fRuleEnsemble; }
149 const MethodBase * GetMethodBase() const { return fMethodBase; }
150
151 private:
152
153 // copy constructor
154 RuleFit( const RuleFit & other );
155
156 // copy method
157 void Copy( const RuleFit & other );
158
159 std::vector<const TMVA::Event *> fTrainingEvents; // all training events
160 std::vector<const TMVA::Event *> fTrainingEventsRndm; // idem, but randomly shuffled
161 std::vector<Double_t> fEventWeights; // original weights of the events - follows fTrainingEvents
162 UInt_t fNTreeSample; // number of events in sub sample = frac*neve
163
164 Double_t fNEveEffTrain; // reweighted number of events = sum(wi)
165 std::vector< const TMVA::DecisionTree *> fForest; // the input forest of decision trees
166 RuleEnsemble fRuleEnsemble; // the ensemble of rules
167 RuleFitParams fRuleFitParams; // fit rule parameters
168 const MethodRuleFit *fMethodRuleFit; // pointer the method which initialized this RuleFit instance
169 const MethodBase *fMethodBase; // pointer the method base which initialized this RuleFit instance
170 Bool_t fVisHistsUseImp; // if true, use importance as weight; else coef in vis hists
171
172 mutable MsgLogger* fLogger; // message logger
173 MsgLogger& Log() const { return *fLogger; }
174
175 static const Int_t randSEED = 0; // set to 1 for debugging purposes or to zero for random seeds
176 std::default_random_engine fRNGEngine;
177
178 ClassDef(RuleFit,0); // Calculations for Friedman's RuleFit method
179 };
180}
181
182#endif
#define d(i)
Definition: RSha256.hxx:102
#define f(i)
Definition: RSha256.hxx:104
#define e(i)
Definition: RSha256.hxx:103
int Int_t
Definition: RtypesCore.h:41
unsigned int UInt_t
Definition: RtypesCore.h:42
const Bool_t kFALSE
Definition: RtypesCore.h:88
bool Bool_t
Definition: RtypesCore.h:59
double Double_t
Definition: RtypesCore.h:55
const Bool_t kTRUE
Definition: RtypesCore.h:87
#define ClassDef(name, id)
Definition: Rtypes.h:324
2-D histogram with a float per channel (see TH1 documentation)}
Definition: TH2.h:250
Implementation of a Decision Tree.
Definition: DecisionTree.h:64
Virtual base Class for all MVA method.
Definition: MethodBase.h:109
J Friedman's RuleFit method.
Definition: MethodRuleFit.h:47
ostringstream derivative to redirect and format output
Definition: MsgLogger.h:59
void SetRuleMinDist(Double_t d)
Definition: RuleEnsemble.h:128
void SetImportanceCut(Double_t minimp=0)
Definition: RuleEnsemble.h:131
A class doing the actual fitting of a linear model using rules as base functions.
Definition: RuleFitParams.h:53
void SetGDPathStep(Double_t s)
Definition: RuleFitParams.h:72
void SetGDTau(Double_t t)
Definition: RuleFitParams.h:86
void SetGDNPathSteps(Int_t np)
Definition: RuleFitParams.h:69
A class implementing various fits of rule ensembles.
Definition: RuleFit.h:45
void GetRndmSampleEvents(std::vector< const TMVA::Event * > &evevec, UInt_t nevents)
draw a random subsample of the training events without replacement
Definition: RuleFit.cxx:467
Double_t EvalEvent(const Event &e)
evaluate single event
Definition: RuleFit.cxx:432
UInt_t fNTreeSample
Definition: RuleFit.h:162
void SetMethodBase(const MethodBase *rfbase)
set MethodBase
Definition: RuleFit.cxx:151
Double_t GetNEveEff() const
Definition: RuleFit.h:131
void InitPtrs(const TMVA::MethodBase *rfbase)
initialize pointers
Definition: RuleFit.cxx:110
void Boost(TMVA::DecisionTree *dt)
Boost the events.
Definition: RuleFit.cxx:339
RuleEnsemble * GetRuleEnsemblePtr()
Definition: RuleFit.h:145
Bool_t fVisHistsUseImp
Definition: RuleFit.h:170
const RuleFitParams & GetRuleFitParams() const
Definition: RuleFit.h:146
void ForestStatistics()
summary of statistics of all trees
Definition: RuleFit.cxx:386
static const Int_t randSEED
Definition: RuleFit.h:175
void CalcImportance()
calculates the importance of each rule
Definition: RuleFit.cxx:418
RuleFitParams * GetRuleFitParamsPtr()
Definition: RuleFit.h:147
void SetMsgType(EMsgType t)
set the current message type to that of mlog for this class and all other subtools
Definition: RuleFit.cxx:191
void Initialize(const TMVA::MethodBase *rfbase)
initialize the parameters of the RuleFit method and make rules
Definition: RuleFit.cxx:120
std::vector< const TMVA::Event * > fTrainingEventsRndm
Definition: RuleFit.h:160
virtual ~RuleFit(void)
destructor
Definition: RuleFit.cxx:90
void FillVisHistCorr(const Rule *rule, std::vector< TH2F * > &hlist)
help routine to MakeVisHists() - fills for all correlation plots
Definition: RuleFit.cxx:715
std::default_random_engine fRNGEngine
Definition: RuleFit.h:176
void InitNEveEff()
init effective number of events (using event weights)
Definition: RuleFit.cxx:98
UInt_t GetNTreeSample() const
Definition: RuleFit.h:130
std::vector< const TMVA::DecisionTree * > fForest
Definition: RuleFit.h:165
void SetGDPathStep(Double_t s=0.01)
Definition: RuleFit.h:114
MsgLogger & Log() const
Definition: RuleFit.h:173
const MethodBase * fMethodBase
Definition: RuleFit.h:169
Double_t GetTrainingEventWeight(UInt_t i) const
Definition: RuleFit.h:133
std::vector< const TMVA::Event * > fTrainingEvents
Definition: RuleFit.h:159
void SaveEventWeights()
save event weights - must be done before making the forest
Definition: RuleFit.cxx:309
void FillCut(TH2F *h2, const TMVA::Rule *rule, Int_t vind)
Fill cut.
Definition: RuleFit.cxx:533
void FillLin(TH2F *h2, Int_t vind)
fill lin
Definition: RuleFit.cxx:584
Bool_t GetCorrVars(TString &title, TString &var1, TString &var2)
get first and second variables from title
Definition: RuleFit.cxx:754
void UseCoefficientsVisHists()
Definition: RuleFit.h:119
const Event * GetTrainingEvent(UInt_t i) const
Definition: RuleFit.h:132
void MakeForest()
make a forest of decisiontrees
Definition: RuleFit.cxx:222
const std::vector< const TMVA::DecisionTree * > & GetForest() const
Definition: RuleFit.h:143
void FitCoefficients()
Fit the coefficients for the rule ensemble.
Definition: RuleFit.cxx:409
void SetRuleMinDist(Double_t d)
Definition: RuleFit.h:111
void SetModelRules()
Definition: RuleFit.h:105
RuleFit(const RuleFit &other)
const MethodRuleFit * fMethodRuleFit
Definition: RuleFit.h:168
const MethodBase * GetMethodBase() const
Definition: RuleFit.h:149
void FillCorr(TH2F *h2, const TMVA::Rule *rule, Int_t v1, Int_t v2)
fill rule correlation between vx and vy, weighted with either the importance or the coefficient
Definition: RuleFit.cxx:608
void NormVisHists(std::vector< TH2F * > &hlist)
normalize rule importance hists
Definition: RuleFit.cxx:486
void SetGDNPathSteps(Int_t n=100)
Definition: RuleFit.h:115
void RestoreEventWeights()
save event weights - must be done before making the forest
Definition: RuleFit.cxx:321
RuleFitParams fRuleFitParams
Definition: RuleFit.h:167
void SetVisHistsUseImp(Bool_t f)
Definition: RuleFit.h:117
void SetModelFull()
Definition: RuleFit.h:107
void MakeVisHists()
this will create histograms visualizing the rule ensemble
Definition: RuleFit.cxx:777
void FillVisHistCut(const Rule *rule, std::vector< TH2F * > &hlist)
help routine to MakeVisHists() - fills for all variables
Definition: RuleFit.cxx:684
std::vector< Double_t > fEventWeights
Definition: RuleFit.h:161
void BuildTree(TMVA::DecisionTree *dt)
build the decision tree using fNTreeSample events from fTrainingEventsRndm
Definition: RuleFit.cxx:201
const std::vector< const TMVA::Event * > & GetTrainingEvents() const
Definition: RuleFit.h:137
void SetGDTau(Double_t t=0.0)
Definition: RuleFit.h:113
const MethodRuleFit * GetMethodRuleFit() const
Definition: RuleFit.h:148
void UseImportanceVisHists()
Definition: RuleFit.h:118
void SetTrainingEvents(const std::vector< const TMVA::Event * > &el)
set the training events randomly
Definition: RuleFit.cxx:440
void ReshuffleEvents()
Definition: RuleFit.h:65
Double_t fNEveEffTrain
Definition: RuleFit.h:164
void SetModelLinear()
Definition: RuleFit.h:103
void Copy(const RuleFit &other)
copy method
Definition: RuleFit.cxx:160
RuleEnsemble fRuleEnsemble
Definition: RuleFit.h:166
const RuleEnsemble & GetRuleEnsemble() const
Definition: RuleFit.h:144
void SetImportanceCut(Double_t minimp=0)
Definition: RuleFit.h:109
Double_t CalcWeightSum(const std::vector< const TMVA::Event * > *events, UInt_t neve=0)
calculate the sum of weights
Definition: RuleFit.cxx:176
RuleFit(void)
default constructor
Definition: RuleFit.cxx:76
MsgLogger * fLogger
Definition: RuleFit.h:172
void MakeDebugHists()
this will create a histograms intended rather for debugging or for the curious user
Definition: RuleFit.cxx:937
Implementation of a rule.
Definition: Rule.h:48
Basic string class.
Definition: TString.h:131
const Int_t n
Definition: legend1.C:16
static constexpr double s
Abstract ClassifierFactory template that handles arbitrary types.