class TMVA::MethodRuleFit: public TMVA::MethodBase
J Friedman's RuleFit method
Function Members (Methods)
public:
protected:
private:
| virtual void | DeclareOptions() |
| virtual void | ProcessOptions() |
Data Members
public:
| enum TMVA::MethodBase::EWeightFileType { | kROOT | |
| kTEXT | ||
| }; | ||
| enum TObject::EStatusBits { | kCanDelete | |
| kMustCleanup | ||
| kObjInCanvas | ||
| kIsReferenced | ||
| kHasUUID | ||
| kCannotPick | ||
| kNoContextMenu | ||
| kInvalidObject | ||
| }; | ||
| enum TObject::[unnamed] { | kIsOnHeap | |
| kNotDeleted | ||
| kZombie | ||
| kBitMask | ||
| kSingleKey | ||
| kOverwrite | ||
| kWriteDelete | ||
| }; |
public:
| Bool_t | TMVA::MethodBase::fSetupCompleted | is method setup |
| const TMVA::Event* | TMVA::MethodBase::fTmpEvent | ! temporary event when testing on a different DataSet than the own one |
protected:
| TMVA::Types::EAnalysisType | TMVA::MethodBase::fAnalysisType | method-mode : true --> regression, false --> classification |
| UInt_t | TMVA::MethodBase::fBackgroundClass | index of the Background-class |
| vector<TString>* | TMVA::MethodBase::fInputVars | vector of input variables used in MVA |
| vector<Float_t>* | TMVA::MethodBase::fMulticlassReturnVal | holds the return-values for the multiclass classification |
| Int_t | TMVA::MethodBase::fNbins | number of bins in input variable histograms |
| Int_t | TMVA::MethodBase::fNbinsH | number of bins in evaluation histograms |
| Int_t | TMVA::MethodBase::fNbinsMVAoutput | number of bins in MVA output histograms |
| TMVA::Ranking* | TMVA::MethodBase::fRanking | pointer to ranking object (created by derived classifiers) |
| vector<Float_t>* | TMVA::MethodBase::fRegressionReturnVal | holds the return-values for the regression |
| UInt_t | TMVA::MethodBase::fSignalClass | index of the Signal-class |
private:
| vector<TMVA::Event*> | fEventSample | the complete training sample |
| vector<TMVA::DecisionTree*,allocator<TMVA::DecisionTree*> > | fForest | the forest |
| TString | fForestTypeS | forest generation: how the trees are generated |
| Double_t | fGDErrScale | GD path: stop |
| Int_t | fGDNPathSteps | GD path: number of steps |
| Double_t | fGDPathEveFrac | GD path: fraction of subsamples used for the fitting |
| Double_t | fGDPathStep | GD path: step size in path |
| Double_t | fGDTau | GD path: def threshhold fraction [0..1] |
| Double_t | fGDTauMax | GD path: max threshhold fraction [0..1] |
| Double_t | fGDTauMin | GD path: min threshhold fraction [0..1] |
| Double_t | fGDTauPrec | GD path: precision of estimated tau |
| UInt_t | fGDTauScan | GD path: number of points to scan |
| Double_t | fGDValidEveFrac | GD path: fraction of subsamples used for the fitting |
| Double_t | fLinQuantile | quantile cut to remove outliers - see RuleEnsemble |
| Double_t | fMaxFracNEve | ditto max |
| Double_t | fMinFracNEve | min fraction of number events |
| Double_t | fMinimp | rule/linear: minimum importance |
| TString | fModelTypeS | rule ensemble: which model (rule,linear or both) |
| TTree* | fMonitorNtuple | pointer to monitor rule ntuple |
| Int_t | fNCuts | grid used in cut applied in node splitting |
| Double_t | fNTCoefficient | ntuple: rule coefficient |
| Double_t | fNTImportance | ntuple: rule importance |
| Int_t | fNTNcuts | ntuple: rule number of cuts |
| Int_t | fNTNvars | ntuple: rule number of vars |
| Double_t | fNTPbb | ntuple: rule P(tag b, true b) |
| Double_t | fNTPbs | ntuple: rule P(tag b, true s) |
| Double_t | fNTPsb | ntuple: rule P(tag s, true b) |
| Double_t | fNTPss | ntuple: rule P(tag s, true s) |
| Double_t | fNTPtag | ntuple: rule P(tag) |
| Double_t | fNTSSB | ntuple: rule S/(S+B) |
| Double_t | fNTSupport | ntuple: rule support |
| Int_t | fNTType | ntuple: rule type (+1->signal, -1->bkg) |
| Int_t | fNTrees | number of trees in forest |
| TMVA::DecisionTree::EPruneMethod | fPruneMethod | forest generation: method used for pruning - see DecisionTree |
| TString | fPruneMethodS | forest generation: prune method - see DecisionTree |
| Double_t | fPruneStrength | forest generation: prune strength - see DecisionTree |
| Int_t | fRFNendnodes | max number of rules (only Friedmans module) |
| Int_t | fRFNrules | max number of rules (only Friedmans module) |
| TString | fRFWorkDir | working directory from Friedmans module |
| TMVA::RuleFit | fRuleFit | RuleFit instance |
| TString | fRuleFitModuleS | which rulefit module to use |
| Double_t | fRuleMinDist | rule min distance - see RuleEnsemble |
| TMVA::SeparationBase* | fSepType | the separation used in node splitting |
| TString | fSepTypeS | forest generation: separation type - see DecisionTree |
| Double_t | fSignalFraction | scalefactor for bkg events to modify initial s/b fraction in training data |
| Double_t | fTreeEveFrac | fraction of events used for traing each tree |
| Bool_t | fUseBoost | use boosted events for forest generation |
| Bool_t | fUseRuleFitJF | if true interface with J.Friedmans RuleFit module |
Class Charts
Function documentation
MethodRuleFit(const TString& jobName, const TString& methodTitle, TMVA::DataSetInfo& theData, const TString& theOption = "", TDirectory* theTargetDir = 0)
standard constructor
MethodRuleFit(TMVA::DataSetInfo& theData, const TString& theWeightFile, TDirectory* theTargetDir = NULL)
constructor from weight file
Bool_t HasAnalysisType(TMVA::Types::EAnalysisType type, UInt_t numberClasses, UInt_t )
RuleFit can handle classification with 2 classes
void DeclareOptions()
define the options (their key words) that can be set in the option string
know options.
general
RuleFitModule <string>
available values are: RFTMVA - use TMVA implementation
RFFriedman - use Friedmans original implementation
Path search (fitting)
GDTau <float> gradient-directed path: fit threshhold, default
GDTauPrec <float> gradient-directed path: precision of estimated tau
GDStep <float> gradient-directed path: step size
GDNSteps <float> gradient-directed path: number of steps
GDErrScale <float> stop scan when error>scale*errmin
Tree generation
fEventsMin <float> minimum fraction of events in a splittable node
fEventsMax <float> maximum fraction of events in a splittable node
nTrees <float> number of trees in forest.
ForestType <string>
available values are: Random - create forest using random subsample and only random variables subset at each node
AdaBoost - create forest with boosted events
Model creation
RuleMinDist <float> min distance allowed between rules
MinImp <float> minimum rule importance accepted
Model <string> model to be used
available values are: ModRuleLinear <default>
ModRule
ModLinear
Friedmans module
RFWorkDir <string> directory where Friedmans module (rf_go.exe) is installed
RFNrules <int> maximum number of rules allowed
RFNendnodes <int> average number of end nodes in the forest of trees
void InitEventSample( void )
write all Events from the Tree into a vector of Events, that are more easily manipulated. This method should never be called without existing trainingTree, as it the vector of events from the ROOT training tree
void Train( void )
void GetHelpMessage() const
get help message text
typical length of text line:
"|--------------------------------------------------------------|"