class TMVA::OptimizeConfigParameters: public TMVA::IFitterTarget

Function Members (Methods)

public:

virtual	~OptimizeConfigParameters()
static TClass*	Class()
TMVA::IFitterTarget	TMVA::IFitterTarget::IFitterTarget()
TMVA::IFitterTarget	TMVA::IFitterTarget::IFitterTarget(const TMVA::IFitterTarget&)
virtual TClass*	IsA() const
TMVA::OptimizeConfigParameters&	operator=(const TMVA::OptimizeConfigParameters&)
map<TString,Double_t>	optimize()
TMVA::OptimizeConfigParameters	OptimizeConfigParameters(const TMVA::OptimizeConfigParameters&)
TMVA::OptimizeConfigParameters	OptimizeConfigParameters(TMVA::MethodBaseconst method, map<TString,TMVA::Interval> tuneParameters, TString fomType = "Separation", TString optimizationType = "GA")
virtual void	TMVA::IFitterTarget::ProgressNotifier(TString, TString)
virtual void	ShowMembers(TMemberInspector& insp) const
virtual void	Streamer(TBuffer&)
void	StreamerNVirtual(TBuffer& ClassDef_StreamerNVirtual_b)

private:

virtual Double_t	EstimatorFunction(vector<Double_t>&)
Double_t	GetBkgEffAtSigEff(Double_t sigEff = 0.5)
Double_t	GetBkgRejAtSigEff(Double_t sigEff = 0.5)
Double_t	GetFOM()
TMVA::MethodBase*	GetMethod()
void	GetMVADists()
Double_t	GetROCIntegral()
vector<int>	GetScanIndices(int val, vector<int> base)
Double_t	GetSeparation()
Double_t	GetSigEffAtBkgEff(Double_t bkgEff = 0.10000000000000001)
TMVA::MsgLogger&	Log() const
void	optimizeFit()
void	optimizeScan()

Data Members

private:

map<vector<Double_t>,Double_t>	fAlreadyTrainedParCombination	save parameters for which the FOM is already known (GA seems to evaluate the same parameters several times)
TString	fFOMType	the FOM type (Separation, ROC integra.. whaeter you implemented..
vector<Float_t>	fFOMvsIter	graph showing the develompment of the Figure Of Merit values during the fit
TMVA::MsgLogger*	fLogger	message logger
TMVA::MethodBase*const	fMethod	The MVA method to be evaluated
TH1D*	fMvaBkg	MVA distrituion for bakgr. events, used for spline fit
TH1D*	fMvaBkgFineBin	MVA distrituion for bakgr. events
TH1D*	fMvaSig	MVA distrituion for signal events, used for spline fit
TH1D*	fMvaSigFineBin	MVA distrituion for signal events
Bool_t	fNotDoneYet	flat to indicate of Method Transformations have been optained yet or not (normally done in MethodBase::TrainMethod)
TString	fOptimizationFitType	which type of optimisation procedure to be used
map<TString,TMVA::Interval*>	fTuneParameters	parameters included in the tuning
map<TString,Double_t>	fTunedParameters	parameters included in the tuning

Class Charts

Inheritance Inherited Members Includes Libraries

Function documentation

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

 Constructor which sets either "Classification or Regression"

~OptimizeConfigParameters()

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

std::map<TString,Double_t> optimize()

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

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

void optimizeScan()

 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

void optimizeFit()

 ranges (intervals) in which the fit varies the parameters

Double_t EstimatorFunction(vector<Double_t>& )

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

Double_t GetFOM()

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

void GetMVADists()

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

Double_t GetSeparation()

 return the searation between the signal and background
 MVA ouput distribution

Double_t GetROCIntegral()

 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!!!

Double_t GetSigEffAtBkgEff(Double_t bkgEff = 0.10000000000000001)

 calculate the signal efficiency for a given background efficiency

Double_t GetBkgEffAtSigEff(Double_t sigEff = 0.5)

 calculate the background efficiency for a given signal efficiency

Double_t GetBkgRejAtSigEff(Double_t sigEff = 0.5)

 calculate the background rejection for a given signal efficiency

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

default constructor

MethodBase* GetMethod()

{return fMethod;}