TMVA::MethodHMatrix

class TMVA::MethodHMatrix : public TMVA::MethodBase

    private:

      Double_t GetChi2(TMVA::Event* e, TMVA::MethodBase::Type) const
          void InitHMatrix()

    public:

                           MethodHMatrix(TString jobName, vector<TString>* theVariables, TTree* theTree = 0, TString theOption = , TDirectory* theTargetDir = 0)
                           MethodHMatrix(vector<TString>* theVariables, TString theWeightFile, TDirectory* theTargetDir = NULL)
                           MethodHMatrix(const TMVA::MethodHMatrix&)
                   virtual ~MethodHMatrix()
            static TClass* Class()
          virtual Double_t GetMvaValue(TMVA::Event* e)
           virtual TClass* IsA() const
      TMVA::MethodHMatrix& operator=(const TMVA::MethodHMatrix&)
              virtual void ReadWeightsFromFile()
              virtual void ShowMembers(TMemberInspector& insp, char* parent)
              virtual void Streamer(TBuffer& b)
                      void StreamerNVirtual(TBuffer& b)
              virtual void Train()
              virtual void WriteHistosToFile()
              virtual void WriteWeightsToFile()

Data Members

    private:

      TMatrixD* fInvHMatrixS         inverse H-matrix (signal)
      TMatrixD* fInvHMatrixB         inverse H-matrix (background)
      TVectorD* fVecMeanS            vector of mean values (signal)
      TVectorD* fVecMeanB            vector of mean values (background)
         Bool_t fNormaliseInputVars  normalise input variables

Class Description

_______________________________________________________________________

H-Matrix method, which is implemented as a simple comparison of chi-squared estimators for signal and background, taking into account the linear correlations between the input variables This MVA approach is used by the DØ collaboration (FNAL) for the purpose of electron identification (see, eg., hep-ex/9507007). As it is implemented in TMVA, it is usually equivalent or worse than the Fisher-Mahalanobis discriminant, and it has only been added for the purpose of completeness. Two χ² estimators are computed for an event, each one for signal and background, using the estimates for the means and covariance matrices obtained from the training sample:
TMVA then uses as normalised analyser for event (i) the ratio: (χ_S(i)² − χ_B²(i)) (χ_S²(i) + χ_B²(i)).

_______________________________________________________________________

 standard constructor for the H-Matrix method

 HMatrix options: none

 constructor to calculate the H-Matrix from the weight file

 default initialisation called by all constructors

 destructor

 computes H-matrices for signal and background samples

 returns the H-matrix signal estimator

 compute chi2-estimator for event according to type (signal/background)

 write matrices and mean vectors to file

 read matrices and mean vectors from file

 write special monitoring histograms to file - not implemented for H-Matrix