root/TMVA/VariableImportance.h

0001 // @(#)root/tmva $Id$
0002 // Author: Omar Zapata and Sergei Gleyzer. 2016
0003
0004
0005 #ifndef ROOT_TMVA_VariableImportance
0006 #define ROOT_TMVA_VariableImportance
0007
0008
0009 #include "TString.h"
0010 #include <vector>
0011
0012 #include "TMVA/Configurable.h"
0013 #include "TMVA/Types.h"
0014
0015 #include <TMVA/Factory.h>
0016
0017 #include <TMVA/DataLoader.h>
0018
0019 #include <TMVA/OptionMap.h>
0020
0021 #include <TMVA/Envelope.h>
0022
0023 namespace TMVA {
0024
0025    class VariableImportanceResult
0026    {
0027      friend class VariableImportance;
0028    private:
0029        OptionMap              fImportanceValues;
0030        std::shared_ptr<TH1F>  fImportanceHist;
0031        VIType                 fType {kShort};
0032    public:
0033        VariableImportanceResult();
0034        VariableImportanceResult(const VariableImportanceResult &);
0035        ~VariableImportanceResult(){fImportanceHist=nullptr;}
0036
0037        OptionMap &GetImportanceValues(){return fImportanceValues;}
0038        TH1F *GetImportanceHist(){return fImportanceHist.get();}
0039        void Print() const ;
0040
0041        TCanvas* Draw(const TString name="VariableImportance") const;
0042    };
0043
0044    class VariableImportance : public Envelope {
0045    private:
0046        UInt_t                    fNumFolds = 0;
0047        VariableImportanceResult  fResults;
0048        VIType                    fType {kShort};
0049    public:
0050        explicit VariableImportance(DataLoader *loader);
0051        ~VariableImportance();
0052
0053        virtual void Evaluate();
0054
0055        void SetType(VIType type){fType=type;}
0056        VIType GetType(){return fType;}
0057
0058        const VariableImportanceResult& GetResults() const {return fResults;}//I need to think about this, which is the best way to get the results?
0059    protected:
0060        //evaluate the simple case that is removing 1 variable at time
0061        void EvaluateImportanceShort();
0062        //evaluate all variables combinations NOTE: use with care in huge datasets with a huge number of variables
0063        void EvaluateImportanceAll();
0064        //evaluate randomly given a number of seeds
0065        void EvaluateImportanceRandom(UInt_t nseeds);
0066
0067        //method to return a nice histogram with the results ;)
0068        TH1F* GetImportance(const UInt_t nbits,std::vector<Float_t> &importances,std::vector<TString> &varNames);
0069
0070        //method to compute the range(number total of operations for every bit configuration)
0071        ULong_t Sum(ULong_t i);
0072
0073    private:
0074        std::unique_ptr<Factory>     fClassifier;
0075        ClassDef(VariableImportance,0);
0076    };
0077 }
0078
0079
0080 #endif