ROOT  6.06/09
Reference Guide
Public Member Functions | Private Member Functions | Private Attributes | List of all members
TMVA::OptimizeConfigParameters Class Reference

Definition at line 63 of file OptimizeConfigParameters.h.

Public Member Functions

 OptimizeConfigParameters (MethodBase *const method, std::map< TString, TMVA::Interval * > tuneParameters, TString fomType="Separation", TString optimizationType="GA")
 
virtual ~OptimizeConfigParameters ()
 the destructor (delete the OptimizeConfigParameters, store the graph and .. delete it) More...
 
std::map< TString, Double_toptimize ()
 
- Public Member Functions inherited from TMVA::IFitterTarget
 IFitterTarget ()
 
virtual ~IFitterTarget ()
 
virtual void ProgressNotifier (TString, TString)
 

Private Member Functions

std::vector< int > GetScanIndices (int val, std::vector< int > base)
 helper function to scan through the all the combinations in the parameter space More...
 
void optimizeScan ()
 do the actual optimization using a simple scan method, i.e. More...
 
void optimizeFit ()
 
Double_t EstimatorFunction (std::vector< Double_t > &)
 return the estimator (from current FOM) for the fitting interface More...
 
Double_t GetFOM ()
 Return the Figure of Merit (FOM) used in the parameter optimization process. More...
 
MethodBaseGetMethod ()
 
void GetMVADists ()
 fill the private histograms with the mva distributinos for sig/bkg More...
 
Double_t GetSeparation ()
 return the searation between the signal and background MVA ouput distribution More...
 
Double_t GetROCIntegral ()
 calculate the area (integral) under the ROC curve as a overall quality measure of the classification More...
 
Double_t GetSigEffAtBkgEff (Double_t bkgEff=0.1)
 calculate the signal efficiency for a given background efficiency More...
 
Double_t GetBkgEffAtSigEff (Double_t sigEff=0.5)
 calculate the background efficiency for a given signal efficiency More...
 
Double_t GetBkgRejAtSigEff (Double_t sigEff=0.5)
 calculate the background rejection for a given signal efficiency More...
 
MsgLoggerLog () const
 

Private Attributes

MethodBase *const fMethod
 
std::vector< Float_tfFOMvsIter
 
std::map< TString, TMVA::Interval * > fTuneParameters
 
std::map< TString, Double_tfTunedParameters
 
std::map< std::vector< Double_t >, Double_tfAlreadyTrainedParCombination
 
TString fFOMType
 
TString fOptimizationFitType
 
TH1DfMvaSig
 
TH1DfMvaBkg
 
TH1DfMvaSigFineBin
 
TH1DfMvaBkgFineBin
 
Bool_t fNotDoneYet
 
MsgLoggerfLogger
 

#include <TMVA/OptimizeConfigParameters.h>

+ Inheritance diagram for TMVA::OptimizeConfigParameters:
+ Collaboration diagram for TMVA::OptimizeConfigParameters:

Constructor & Destructor Documentation

TMVA::OptimizeConfigParameters::OptimizeConfigParameters ( MethodBase *const  method,
std::map< TString, TMVA::Interval * >  tuneParameters,
TString  fomType = "Separation",
TString  optimizationType = "GA" 
)
TMVA::OptimizeConfigParameters::~OptimizeConfigParameters ( )
virtual

the destructor (delete the OptimizeConfigParameters, store the graph and .. delete it)

Definition at line 86 of file OptimizeConfigParameters.cxx.

Member Function Documentation

Double_t TMVA::OptimizeConfigParameters::EstimatorFunction ( std::vector< Double_t > &  pars)
privatevirtual

return the estimator (from current FOM) for the fitting interface

Implements TMVA::IFitterTarget.

Definition at line 290 of file OptimizeConfigParameters.cxx.

Double_t TMVA::OptimizeConfigParameters::GetBkgEffAtSigEff ( Double_t  sigEff = 0.5)
private

calculate the background efficiency for a given signal efficiency

Definition at line 516 of file OptimizeConfigParameters.cxx.

Double_t TMVA::OptimizeConfigParameters::GetBkgRejAtSigEff ( Double_t  sigEff = 0.5)
private

calculate the background rejection for a given signal efficiency

Definition at line 552 of file OptimizeConfigParameters.cxx.

Double_t TMVA::OptimizeConfigParameters::GetFOM ( )
private

Return the Figure of Merit (FOM) used in the parameter optimization process.

Definition at line 333 of file OptimizeConfigParameters.cxx.

MethodBase* TMVA::OptimizeConfigParameters::GetMethod ( )
inlineprivate

Definition at line 85 of file OptimizeConfigParameters.h.

Referenced by ~OptimizeConfigParameters().

void TMVA::OptimizeConfigParameters::GetMVADists ( )
private

fill the private histograms with the mva distributinos for sig/bkg

Definition at line 362 of file OptimizeConfigParameters.cxx.

Double_t TMVA::OptimizeConfigParameters::GetROCIntegral ( )
private

calculate the area (integral) under the ROC curve as a overall quality measure of the classification

makeing pdfs out of the MVA-ouput distributions doesn't work reliably for cases where the MVA-ouput isn't a smooth distribution. this happens "frequently" in BDTs for example when the number of trees is small resulting in only some discrete possible MVA ouput values. (I still leave the code here, but use this with care!!! The default however is to use the distributions!!!

Definition at line 428 of file OptimizeConfigParameters.cxx.

std::vector< int > TMVA::OptimizeConfigParameters::GetScanIndices ( int  val,
std::vector< int >  base 
)
private

helper function to scan through the all the combinations in the parameter space

Definition at line 138 of file OptimizeConfigParameters.cxx.

Double_t TMVA::OptimizeConfigParameters::GetSeparation ( )
private

return the searation between the signal and background MVA ouput distribution

Definition at line 403 of file OptimizeConfigParameters.cxx.

Double_t TMVA::OptimizeConfigParameters::GetSigEffAtBkgEff ( Double_t  bkgEff = 0.1)
private

calculate the signal efficiency for a given background efficiency

Definition at line 480 of file OptimizeConfigParameters.cxx.

MsgLogger& TMVA::OptimizeConfigParameters::Log ( ) const
inlineprivate

Definition at line 111 of file OptimizeConfigParameters.h.

std::map< TString, Double_t > TMVA::OptimizeConfigParameters::optimize ( )
void TMVA::OptimizeConfigParameters::optimizeFit ( )
private

Definition at line 229 of file OptimizeConfigParameters.cxx.

void TMVA::OptimizeConfigParameters::optimizeScan ( )
private

do the actual optimization using a simple scan method, i.e.

calcualte the FOM for different tuning paraemters and remember which one is gave the best FOM

Definition at line 153 of file OptimizeConfigParameters.cxx.

Member Data Documentation

std::map< std::vector<Double_t> , Double_t> TMVA::OptimizeConfigParameters::fAlreadyTrainedParCombination
private

Definition at line 99 of file OptimizeConfigParameters.h.

TString TMVA::OptimizeConfigParameters::fFOMType
private

Definition at line 100 of file OptimizeConfigParameters.h.

Referenced by ~OptimizeConfigParameters().

std::vector<Float_t> TMVA::OptimizeConfigParameters::fFOMvsIter
private

Definition at line 96 of file OptimizeConfigParameters.h.

Referenced by ~OptimizeConfigParameters().

MsgLogger* TMVA::OptimizeConfigParameters::fLogger
mutableprivate

Definition at line 110 of file OptimizeConfigParameters.h.

Referenced by Log().

MethodBase* const TMVA::OptimizeConfigParameters::fMethod
private

Definition at line 95 of file OptimizeConfigParameters.h.

Referenced by GetMethod().

TH1D* TMVA::OptimizeConfigParameters::fMvaBkg
private

Definition at line 103 of file OptimizeConfigParameters.h.

TH1D* TMVA::OptimizeConfigParameters::fMvaBkgFineBin
private

Definition at line 106 of file OptimizeConfigParameters.h.

TH1D* TMVA::OptimizeConfigParameters::fMvaSig
private

Definition at line 102 of file OptimizeConfigParameters.h.

TH1D* TMVA::OptimizeConfigParameters::fMvaSigFineBin
private

Definition at line 105 of file OptimizeConfigParameters.h.

Bool_t TMVA::OptimizeConfigParameters::fNotDoneYet
private

Definition at line 108 of file OptimizeConfigParameters.h.

TString TMVA::OptimizeConfigParameters::fOptimizationFitType
private

Definition at line 101 of file OptimizeConfigParameters.h.

Referenced by ~OptimizeConfigParameters().

std::map<TString,Double_t> TMVA::OptimizeConfigParameters::fTunedParameters
private

Definition at line 98 of file OptimizeConfigParameters.h.

std::map<TString,TMVA::Interval*> TMVA::OptimizeConfigParameters::fTuneParameters
private

Definition at line 97 of file OptimizeConfigParameters.h.


The documentation for this class was generated from the following files: