Analysis of Boosted Decision Trees.

Boosted decision trees have been successfully used in High Energy Physics analysis for example by the MiniBooNE experiment (Yang-Roe-Zhu, physics/0508045). In Boosted Decision Trees, the selection is done on a majority vote on the result of several decision trees, which are all derived from the same training sample by supplying different event weights during the training.

Decision trees:

Successive decision nodes are used to categorize the events out of the sample as either signal or background. Each node uses only a single discriminating variable to decide if the event is signal-like ("goes right") or background-like ("goes left"). This forms a tree like structure with "baskets" at the end (leave nodes), and an event is classified as either signal or background according to whether the basket where it ends up has been classified signal or background during the training. Training of a decision tree is the process to define the "cut criteria" for each node. The training starts with the root node. Here one takes the full training event sample and selects the variable and corresponding cut value that gives the best separation between signal and background at this stage. Using this cut criterion, the sample is then divided into two subsamples, a signal-like (right) and a background-like (left) sample. Two new nodes are then created for each of the two sub-samples and they are constructed using the same mechanism as described for the root node. The devision is stopped once a certain node has reached either a minimum number of events, or a minimum or maximum signal purity. These leave nodes are then called "signal" or "background" if they contain more signal respective background events from the training sample.

Boosting:

The idea behind adaptive boosting (AdaBoost) is, that signal events from the training sample, that end up in a background node (and vice versa) are given a larger weight than events that are in the correct leave node. This results in a re-weighed training event sample, with which then a new decision tree can be developed. The boosting can be applied several times (typically 100-500 times) and one ends up with a set of decision trees (a forest). Gradient boosting works more like a function expansion approach, where each tree corresponds to a summand. The parameters for each summand (tree) are determined by the minimization of a error function (binomial log- likelihood for classification and Huber loss for regression). A greedy algorithm is used, which means, that only one tree is modified at a time, while the other trees stay fixed.

Bagging:

In this particular variant of the Boosted Decision Trees the boosting is not done on the basis of previous training results, but by a simple stochastic re-sampling of the initial training event sample.

Random Trees:

Similar to the "Random Forests" from Leo Breiman and Adele Cutler, it uses the bagging algorithm together and bases the determination of the best node-split during the training on a random subset of variables only which is individually chosen for each split.

Analysis:

Applying an individual decision tree to a test event results in a classification of the event as either signal or background. For the boosted decision tree selection, an event is successively subjected to the whole set of decision trees and depending on how often it is classified as signal, a "likelihood" estimator is constructed for the event being signal or background. The value of this estimator is the one which is then used to select the events from an event sample, and the cut value on this estimator defines the efficiency and purity of the selection.

Definition at line 63 of file MethodBDT.h.

Public Member Functions
	MethodBDT (const TString &jobName, const TString &methodTitle, DataSetInfo &theData, const TString &theOption="")
	The standard constructor for the "boosted decision trees".

	MethodBDT (DataSetInfo &theData, const TString &theWeightFile)

virtual	~MethodBDT (void)
	Destructor.

void	AddWeightsXMLTo (void *parent) const
	Write weights to XML.

Double_t	Boost (std::vector< const TMVA::Event * > &, DecisionTree *dt, UInt_t cls=0)
	Apply the boosting algorithm (the algorithm is selecte via the "option" given in the constructor.

const Ranking *	CreateRanking ()
	Compute ranking of input variables.

void	DeclareOptions ()
	Define the options (their key words).

const std::vector< double > &	GetBoostWeights () const

const std::vector< TMVA::DecisionTree * > &	GetForest () const

void	GetHelpMessage () const
	Get help message text.

const std::vector< Float_t > &	GetMulticlassValues ()
	Get the multiclass MVA response for the BDT classifier.

Double_t	GetMvaValue (Double_t err=nullptr, Double_t errUpper=nullptr)

UInt_t	GetNTrees () const

const std::vector< Float_t > &	GetRegressionValues ()
	Get the regression value generated by the BDTs.

const std::vector< const TMVA::Event * > &	GetTrainingEvents () const

std::vector< Double_t >	GetVariableImportance ()
	Return the relative variable importance, normalized to all variables together having the importance 1.

Double_t	GetVariableImportance (UInt_t ivar)
	Returns the measure for the variable importance of variable "ivar" which is later used in GetVariableImportance() to calculate the relative variable importances.

virtual Bool_t	HasAnalysisType (Types::EAnalysisType type, UInt_t numberClasses, UInt_t numberTargets)
	BDT can handle classification with multiple classes and regression with one regression-target.

void	InitEventSample ()
	Initialize the event sample (i.e. reset the boost-weights... etc).

virtual TClass *	IsA () const

void	MakeClassInstantiateNode (DecisionTreeNode *n, std::ostream &fout, const TString &className) const
	Recursively descends a tree and writes the node instance to the output stream.

void	MakeClassSpecific (std::ostream &, const TString &) const
	Make ROOT-independent C++ class for classifier response (classifier-specific implementation).

void	MakeClassSpecificHeader (std::ostream &, const TString &) const
	Specific class header.

virtual std::map< TString, Double_t >	OptimizeTuningParameters (TString fomType="ROCIntegral", TString fitType="FitGA")
	Call the Optimizer with the set of parameters and ranges that are meant to be tuned.

void	ProcessOptions ()
	The option string is decoded, for available options see "DeclareOptions".

virtual void	ReadWeightsFromStream (std::istream &)=0

void	ReadWeightsFromStream (std::istream &istr)
	Read the weights (BDT coefficients).

virtual void	ReadWeightsFromStream (TFile &)

void	ReadWeightsFromXML (void *parent)
	Reads the BDT from the xml file.

void	Reset (void)
	Reset the method, as if it had just been instantiated (forget all training etc.).

void	SetAdaBoostBeta (Double_t b)

void	SetBaggedSampleFraction (Double_t f)

void	SetMaxDepth (Int_t d)

void	SetMinNodeSize (Double_t sizeInPercent)

void	SetMinNodeSize (TString sizeInPercent)

void	SetNodePurityLimit (Double_t l)

void	SetNTrees (Int_t d)

void	SetShrinkage (Double_t s)

virtual void	SetTuneParameters (std::map< TString, Double_t > tuneParameters)
	Set the tuning parameters according to the argument.

void	SetUseNvars (Int_t n)

virtual void	Streamer (TBuffer &)

void	StreamerNVirtual (TBuffer &ClassDef_StreamerNVirtual_b)

Double_t	TestTreeQuality (DecisionTree *dt)
	Test the tree quality.. in terms of Misclassification.

void	Train (void)
	BDT training.

void	WriteMonitoringHistosToFile (void) const
	Here we could write some histograms created during the processing to the output file.

Public Member Functions inherited from TMVA::MethodBase
	MethodBase (const TString &jobName, Types::EMVA methodType, const TString &methodTitle, DataSetInfo &dsi, const TString &theOption="")
	standard constructor

	MethodBase (Types::EMVA methodType, DataSetInfo &dsi, const TString &weightFile)
	constructor used for Testing + Application of the MVA, only (no training), using given WeightFiles

virtual	~MethodBase ()
	destructor

void	AddOutput (Types::ETreeType type, Types::EAnalysisType analysisType)

TDirectory *	BaseDir () const
	returns the ROOT directory where info/histograms etc of the corresponding MVA method instance are stored

virtual void	CheckSetup ()
	check may be overridden by derived class (sometimes, eg, fitters are used which can only be implemented during training phase)

DataSet *	Data () const

DataSetInfo &	DataInfo () const

void	DisableWriting (Bool_t setter)

Bool_t	DoMulticlass () const

Bool_t	DoRegression () const

void	ExitFromTraining ()

Types::EAnalysisType	GetAnalysisType () const

UInt_t	GetCurrentIter ()

virtual Double_t	GetEfficiency (const TString &, Types::ETreeType, Double_t &err)
	fill background efficiency (resp.

const Event *	GetEvent () const

const Event *	GetEvent (const TMVA::Event *ev) const

const Event *	GetEvent (Long64_t ievt) const

const Event *	GetEvent (Long64_t ievt, Types::ETreeType type) const

const std::vector< TMVA::Event * > &	GetEventCollection (Types::ETreeType type)
	returns the event collection (i.e.

TFile *	GetFile () const

const TString &	GetInputLabel (Int_t i) const

const char *	GetInputTitle (Int_t i) const

const TString &	GetInputVar (Int_t i) const

TMultiGraph *	GetInteractiveTrainingError ()

const TString &	GetJobName () const

virtual Double_t	GetKSTrainingVsTest (Char_t SorB, TString opt="X")

virtual Double_t	GetMaximumSignificance (Double_t SignalEvents, Double_t BackgroundEvents, Double_t &optimal_significance_value) const
	plot significance, \( \frac{S}{\sqrt{S^2 + B^2}} \), curve for given number of signal and background events; returns cut for maximum significance also returned via reference is the maximum significance

UInt_t	GetMaxIter ()

Double_t	GetMean (Int_t ivar) const

const TString &	GetMethodName () const

Types::EMVA	GetMethodType () const

TString	GetMethodTypeName () const

virtual TMatrixD	GetMulticlassConfusionMatrix (Double_t effB, Types::ETreeType type)
	Construct a confusion matrix for a multiclass classifier.

virtual std::vector< Float_t >	GetMulticlassEfficiency (std::vector< std::vector< Float_t > > &purity)

virtual std::vector< Float_t >	GetMulticlassTrainingEfficiency (std::vector< std::vector< Float_t > > &purity)

Double_t	GetMvaValue (const TMVA::Event const ev, Double_t err=nullptr, Double_t *errUpper=nullptr)

const char *	GetName () const

UInt_t	GetNEvents () const

UInt_t	GetNTargets () const

UInt_t	GetNvar () const

UInt_t	GetNVariables () const

virtual Double_t	GetProba (const Event *ev)

virtual Double_t	GetProba (Double_t mvaVal, Double_t ap_sig)
	compute likelihood ratio

const TString	GetProbaName () const

virtual Double_t	GetRarity (Double_t mvaVal, Types::ESBType reftype=Types::kBackground) const
	compute rarity:

virtual void	GetRegressionDeviation (UInt_t tgtNum, Types::ETreeType type, Double_t &stddev, Double_t &stddev90Percent) const

const std::vector< Float_t > &	GetRegressionValues (const TMVA::Event *const ev)

Double_t	GetRMS (Int_t ivar) const

virtual Double_t	GetROCIntegral (PDF pdfS=nullptr, PDF pdfB=nullptr) const
	calculate the area (integral) under the ROC curve as a overall quality measure of the classification

virtual Double_t	GetROCIntegral (TH1D histS, TH1D histB) const
	calculate the area (integral) under the ROC curve as a overall quality measure of the classification

virtual Double_t	GetSeparation (PDF pdfS=nullptr, PDF pdfB=nullptr) const
	compute "separation" defined as

virtual Double_t	GetSeparation (TH1 , TH1 ) const
	compute "separation" defined as

Double_t	GetSignalReferenceCut () const

Double_t	GetSignalReferenceCutOrientation () const

virtual Double_t	GetSignificance () const
	compute significance of mean difference

const Event *	GetTestingEvent (Long64_t ievt) const

Double_t	GetTestTime () const

const TString &	GetTestvarName () const

virtual Double_t	GetTrainingEfficiency (const TString &)

const Event *	GetTrainingEvent (Long64_t ievt) const

virtual const std::vector< Float_t > &	GetTrainingHistory (const char *)

UInt_t	GetTrainingROOTVersionCode () const

TString	GetTrainingROOTVersionString () const
	calculates the ROOT version string from the training version code on the fly

UInt_t	GetTrainingTMVAVersionCode () const

TString	GetTrainingTMVAVersionString () const
	calculates the TMVA version string from the training version code on the fly

Double_t	GetTrainTime () const

TransformationHandler &	GetTransformationHandler (Bool_t takeReroutedIfAvailable=true)

const TransformationHandler &	GetTransformationHandler (Bool_t takeReroutedIfAvailable=true) const

TString	GetWeightFileName () const
	retrieve weight file name

Double_t	GetXmax (Int_t ivar) const

Double_t	GetXmin (Int_t ivar) const

Bool_t	HasMVAPdfs () const

void	InitIPythonInteractive ()

Bool_t	IsModelPersistence () const

virtual Bool_t	IsSignalLike ()
	uses a pre-set cut on the MVA output (SetSignalReferenceCut and SetSignalReferenceCutOrientation) for a quick determination if an event would be selected as signal or background

virtual Bool_t	IsSignalLike (Double_t mvaVal)
	uses a pre-set cut on the MVA output (SetSignalReferenceCut and SetSignalReferenceCutOrientation) for a quick determination if an event with this mva output value would be selected as signal or background

Bool_t	IsSilentFile () const

virtual void	MakeClass (const TString &classFileName=TString("")) const
	create reader class for method (classification only at present)

TDirectory *	MethodBaseDir () const
	returns the ROOT directory where all instances of the corresponding MVA method are stored

void	PrintHelpMessage () const
	prints out method-specific help method

void	ProcessSetup ()
	process all options the "CheckForUnusedOptions" is done in an independent call, since it may be overridden by derived class (sometimes, eg, fitters are used which can only be implemented during training phase)

void	ReadStateFromFile ()
	Function to write options and weights to file.

void	ReadStateFromStream (std::istream &tf)
	read the header from the weight files of the different MVA methods

void	ReadStateFromStream (TFile &rf)
	write reference MVA distributions (and other information) to a ROOT type weight file

void	ReadStateFromXMLString (const char *xmlstr)
	for reading from memory

void	RerouteTransformationHandler (TransformationHandler *fTargetTransformation)

virtual void	SetAnalysisType (Types::EAnalysisType type)

void	SetBaseDir (TDirectory *methodDir)

void	SetFile (TFile *file)

void	SetMethodBaseDir (TDirectory *methodDir)

void	SetMethodDir (TDirectory *methodDir)

void	SetModelPersistence (Bool_t status)

void	SetSignalReferenceCut (Double_t cut)

void	SetSignalReferenceCutOrientation (Double_t cutOrientation)

void	SetSilentFile (Bool_t status)

void	SetTestTime (Double_t testTime)

void	SetTestvarName (const TString &v="")

void	SetTrainTime (Double_t trainTime)

void	SetupMethod ()
	setup of methods

void	StreamerNVirtual (TBuffer &ClassDef_StreamerNVirtual_b)

virtual void	TestClassification ()
	initialization

virtual void	TestMulticlass ()
	test multiclass classification

virtual void	TestRegression (Double_t &bias, Double_t &biasT, Double_t &dev, Double_t &devT, Double_t &rms, Double_t &rmsT, Double_t &mInf, Double_t &mInfT, Double_t &corr, Types::ETreeType type)
	calculate <sum-of-deviation-squared> of regression output versus "true" value from test sample

bool	TrainingEnded ()

void	TrainMethod ()

virtual void	WriteEvaluationHistosToFile (Types::ETreeType treetype)
	writes all MVA evaluation histograms to file

void	WriteStateToFile () const
	write options and weights to file note that each one text file for the main configuration information and one ROOT file for ROOT objects are created

Public Member Functions inherited from TMVA::IMethod
	IMethod ()

virtual	~IMethod ()

void	StreamerNVirtual (TBuffer &ClassDef_StreamerNVirtual_b)

Public Member Functions inherited from TMVA::Configurable
	Configurable (const TString &theOption="")
	constructor

virtual	~Configurable ()
	default destructor

void	AddOptionsXMLTo (void *parent) const
	write options to XML file

template<class T >
void	AddPreDefVal (const T &)

template<class T >
void	AddPreDefVal (const TString &optname, const T &)

void	CheckForUnusedOptions () const
	checks for unused options in option string

template<class T >
TMVA::OptionBase *	DeclareOptionRef (T &ref, const TString &name, const TString &desc)

template<class T >
OptionBase *	DeclareOptionRef (T &ref, const TString &name, const TString &desc="")

template<class T >
TMVA::OptionBase *	DeclareOptionRef (T *&ref, Int_t size, const TString &name, const TString &desc)

template<class T >
OptionBase *	DeclareOptionRef (T *&ref, Int_t size, const TString &name, const TString &desc="")

const char *	GetConfigDescription () const

const char *	GetConfigName () const

const TString &	GetOptions () const

MsgLogger &	Log () const

virtual void	ParseOptions ()
	options parser

void	PrintOptions () const
	prints out the options set in the options string and the defaults

void	ReadOptionsFromStream (std::istream &istr)
	read option back from the weight file

void	ReadOptionsFromXML (void *node)

void	SetConfigDescription (const char *d)

void	SetConfigName (const char *n)

void	SetMsgType (EMsgType t)

void	SetOptions (const TString &s)

void	StreamerNVirtual (TBuffer &ClassDef_StreamerNVirtual_b)

void	WriteOptionsToStream (std::ostream &o, const TString &prefix) const
	write options to output stream (e.g. in writing the MVA weight files

Public Member Functions inherited from TNamed
	TNamed ()

	TNamed (const char name, const char title)

	TNamed (const TNamed &named)
	TNamed copy ctor.

	TNamed (const TString &name, const TString &title)

virtual	~TNamed ()
	TNamed destructor.

void	Clear (Option_t *option="") override
	Set name and title to empty strings ("").

TObject *	Clone (const char *newname="") const override
	Make a clone of an object using the Streamer facility.

Int_t	Compare (const TObject *obj) const override
	Compare two TNamed objects.

void	Copy (TObject &named) const override
	Copy this to obj.

virtual void	FillBuffer (char *&buffer)
	Encode TNamed into output buffer.

const char *	GetName () const override
	Returns name of object.

const char *	GetTitle () const override
	Returns title of object.

ULong_t	Hash () const override
	Return hash value for this object.

TClass *	IsA () const override

Bool_t	IsSortable () const override

void	ls (Option_t *option="") const override
	List TNamed name and title.

TNamed &	operator= (const TNamed &rhs)
	TNamed assignment operator.

void	Print (Option_t *option="") const override
	Print TNamed name and title.

virtual void	SetName (const char *name)
	Set the name of the TNamed.

virtual void	SetNameTitle (const char name, const char title)
	Set all the TNamed parameters (name and title).

virtual void	SetTitle (const char *title="")
	Set the title of the TNamed.

virtual Int_t	Sizeof () const
	Return size of the TNamed part of the TObject.

void	Streamer (TBuffer &) override
	Stream an object of class TObject.

void	StreamerNVirtual (TBuffer &ClassDef_StreamerNVirtual_b)

Public Member Functions inherited from TObject
	TObject ()
	TObject constructor.

	TObject (const TObject &object)
	TObject copy ctor.

virtual	~TObject ()
	TObject destructor.

void	AbstractMethod (const char *method) const
	Use this method to implement an "abstract" method that you don't want to leave purely abstract.

virtual void	AppendPad (Option_t *option="")
	Append graphics object to current pad.

virtual void	Browse (TBrowser *b)
	Browse object. May be overridden for another default action.

ULong_t	CheckedHash ()
	Check and record whether this class has a consistent Hash/RecursiveRemove setup (*) and then return the regular Hash value for this object.

virtual const char *	ClassName () const
	Returns name of class to which the object belongs.

virtual void	Delete (Option_t *option="")
	Delete this object.

virtual Int_t	DistancetoPrimitive (Int_t px, Int_t py)
	Computes distance from point (px,py) to the object.

virtual void	Draw (Option_t *option="")
	Default Draw method for all objects.

virtual void	DrawClass () const
	Draw class inheritance tree of the class to which this object belongs.

virtual TObject *	DrawClone (Option_t *option="") const
	Draw a clone of this object in the current selected pad with: `gROOT->SetSelectedPad(c1)`.

virtual void	Dump () const
	Dump contents of object on stdout.

virtual void	Error (const char method, const char msgfmt,...) const
	Issue error message.

virtual void	Execute (const char method, const char params, Int_t *error=nullptr)
	Execute method on this object with the given parameter string, e.g.

virtual void	Execute (TMethod method, TObjArray params, Int_t *error=nullptr)
	Execute method on this object with parameters stored in the TObjArray.

virtual void	ExecuteEvent (Int_t event, Int_t px, Int_t py)
	Execute action corresponding to an event at (px,py).

virtual void	Fatal (const char method, const char msgfmt,...) const
	Issue fatal error message.

virtual TObject *	FindObject (const char *name) const
	Must be redefined in derived classes.

virtual TObject *	FindObject (const TObject *obj) const
	Must be redefined in derived classes.

virtual Option_t *	GetDrawOption () const
	Get option used by the graphics system to draw this object.

virtual const char *	GetIconName () const
	Returns mime type name of object.

virtual char *	GetObjectInfo (Int_t px, Int_t py) const
	Returns string containing info about the object at position (px,py).

virtual Option_t *	GetOption () const

virtual UInt_t	GetUniqueID () const
	Return the unique object id.

virtual Bool_t	HandleTimer (TTimer *timer)
	Execute action in response of a timer timing out.

Bool_t	HasInconsistentHash () const
	Return true is the type of this object is known to have an inconsistent setup for Hash and RecursiveRemove (i.e.

virtual void	Info (const char method, const char msgfmt,...) const
	Issue info message.

virtual Bool_t	InheritsFrom (const char *classname) const
	Returns kTRUE if object inherits from class "classname".

virtual Bool_t	InheritsFrom (const TClass *cl) const
	Returns kTRUE if object inherits from TClass cl.

virtual void	Inspect () const
	Dump contents of this object in a graphics canvas.

void	InvertBit (UInt_t f)

Bool_t	IsDestructed () const
	IsDestructed.

virtual Bool_t	IsEqual (const TObject *obj) const
	Default equal comparison (objects are equal if they have the same address in memory).

virtual Bool_t	IsFolder () const
	Returns kTRUE in case object contains browsable objects (like containers or lists of other objects).

R__ALWAYS_INLINE Bool_t	IsOnHeap () const

R__ALWAYS_INLINE Bool_t	IsZombie () const

void	MayNotUse (const char *method) const
	Use this method to signal that a method (defined in a base class) may not be called in a derived class (in principle against good design since a child class should not provide less functionality than its parent, however, sometimes it is necessary).

virtual Bool_t	Notify ()
	This method must be overridden to handle object notification (the base implementation is no-op).

void	Obsolete (const char method, const char asOfVers, const char *removedFromVers) const
	Use this method to declare a method obsolete.

void	operator delete (void *ptr)
	Operator delete.

void	operator delete (void ptr, void vp)
	Only called by placement new when throwing an exception.

void	operator delete[] (void *ptr)
	Operator delete [].

void	operator delete[] (void ptr, void vp)
	Only called by placement new[] when throwing an exception.

void *	operator new (size_t sz)

void *	operator new (size_t sz, void *vp)

void *	operator new[] (size_t sz)

void *	operator new[] (size_t sz, void *vp)

TObject &	operator= (const TObject &rhs)
	TObject assignment operator.

virtual void	Paint (Option_t *option="")
	This method must be overridden if a class wants to paint itself.

virtual void	Pop ()
	Pop on object drawn in a pad to the top of the display list.

virtual Int_t	Read (const char *name)
	Read contents of object with specified name from the current directory.

virtual void	RecursiveRemove (TObject *obj)
	Recursively remove this object from a list.

void	ResetBit (UInt_t f)

virtual void	SaveAs (const char filename="", Option_t option="") const
	Save this object in the file specified by filename.

virtual void	SavePrimitive (std::ostream &out, Option_t *option="")
	Save a primitive as a C++ statement(s) on output stream "out".

void	SetBit (UInt_t f)

void	SetBit (UInt_t f, Bool_t set)
	Set or unset the user status bits as specified in f.

virtual void	SetDrawOption (Option_t *option="")
	Set drawing option for object.

virtual void	SetUniqueID (UInt_t uid)
	Set the unique object id.

void	StreamerNVirtual (TBuffer &ClassDef_StreamerNVirtual_b)

virtual void	SysError (const char method, const char msgfmt,...) const
	Issue system error message.

R__ALWAYS_INLINE Bool_t	TestBit (UInt_t f) const

Int_t	TestBits (UInt_t f) const

virtual void	UseCurrentStyle ()
	Set current style settings in this object This function is called when either TCanvas::UseCurrentStyle or TROOT::ForceStyle have been invoked.

virtual void	Warning (const char method, const char msgfmt,...) const
	Issue warning message.

virtual Int_t	Write (const char *name=nullptr, Int_t option=0, Int_t bufsize=0)
	Write this object to the current directory.

virtual Int_t	Write (const char *name=nullptr, Int_t option=0, Int_t bufsize=0) const
	Write this object to the current directory.

Static Public Member Functions
static TClass *	Class ()

static const char *	Class_Name ()

static constexpr Version_t	Class_Version ()

static const char *	DeclFileName ()

Static Public Member Functions inherited from TMVA::MethodBase
static TClass *	Class ()

static const char *	Class_Name ()

static constexpr Version_t	Class_Version ()

static const char *	DeclFileName ()

Static Public Member Functions inherited from TMVA::IMethod
static TClass *	Class ()

static const char *	Class_Name ()

static constexpr Version_t	Class_Version ()

static const char *	DeclFileName ()

Static Public Member Functions inherited from TMVA::Configurable
static TClass *	Class ()

static const char *	Class_Name ()

static constexpr Version_t	Class_Version ()

static const char *	DeclFileName ()

Static Public Member Functions inherited from TNamed
static TClass *	Class ()

static const char *	Class_Name ()

static constexpr Version_t	Class_Version ()

static const char *	DeclFileName ()

Static Public Member Functions inherited from TObject
static TClass *	Class ()

static const char *	Class_Name ()

static constexpr Version_t	Class_Version ()

static const char *	DeclFileName ()

static Longptr_t	GetDtorOnly ()
	Return destructor only flag.

static Bool_t	GetObjectStat ()
	Get status of object stat flag.

static void	SetDtorOnly (void *obj)
	Set destructor only flag.

static void	SetObjectStat (Bool_t stat)
	Turn on/off tracking of objects in the TObjectTable.

Protected Member Functions
void	DeclareCompatibilityOptions ()
	Options that are used ONLY for the READER to ensure backward compatibility.

Protected Member Functions inherited from TMVA::MethodBase
virtual std::vector< Double_t >	GetDataMvaValues (DataSet *data=nullptr, Long64_t firstEvt=0, Long64_t lastEvt=-1, Bool_t logProgress=false)
	get all the MVA values for the events of the given Data type

const TString &	GetInternalVarName (Int_t ivar) const

virtual std::vector< Double_t >	GetMvaValues (Long64_t firstEvt=0, Long64_t lastEvt=-1, Bool_t logProgress=false)
	get all the MVA values for the events of the current Data type

const TString &	GetOriginalVarName (Int_t ivar) const

const TString &	GetWeightFileDir () const

Bool_t	HasTrainingTree () const

Bool_t	Help () const

Bool_t	IgnoreEventsWithNegWeightsInTraining () const

Bool_t	IsConstructedFromWeightFile () const

Bool_t	IsNormalised () const

void	NoErrorCalc (Double_t const err, Double_t const errUpper)

void	SetNormalised (Bool_t norm)

void	SetWeightFileDir (TString fileDir)
	set directory of weight file

void	SetWeightFileName (TString)
	set the weight file name (depreciated)

void	Statistics (Types::ETreeType treeType, const TString &theVarName, Double_t &, Double_t &, Double_t &, Double_t &, Double_t &, Double_t &)
	calculates rms,mean, xmin, xmax of the event variable this can be either done for the variables as they are or for normalised variables (in the range of 0-1) if "norm" is set to kTRUE

Bool_t	TxtWeightsOnly () const

Bool_t	Verbose () const

Protected Member Functions inherited from TMVA::Configurable
void	EnableLooseOptions (Bool_t b=kTRUE)

const TString &	GetReferenceFile () const

Bool_t	LooseOptionCheckingEnabled () const

void	ResetSetFlag ()
	resets the IsSet flag for all declare options to be called before options are read from stream

void	WriteOptionsReferenceToFile ()
	write complete options to output stream

Protected Member Functions inherited from TObject
virtual void	DoError (int level, const char location, const char fmt, va_list va) const
	Interface to ErrorHandler (protected).

void	MakeZombie ()

Private Member Functions
Double_t	AdaBoost (std::vector< const TMVA::Event * > &, DecisionTree *dt)
	The AdaBoost implementation.

Double_t	AdaBoostR2 (std::vector< const TMVA::Event * > &, DecisionTree *dt)
	Adaption of the AdaBoost to regression problems (see H.Drucker 1997).

Double_t	AdaCost (std::vector< const TMVA::Event * > &, DecisionTree *dt)
	The AdaCost boosting algorithm takes a simple cost Matrix (currently fixed for all events... later could be modified to use individual cost matrices for each events as in the original paper...

Double_t	ApplyPreselectionCuts (const Event *ev)
	Apply the preselection cuts before even bothering about any Decision Trees in the GetMVA .

Double_t	Bagging ()
	Call it boot-strapping, re-sampling or whatever you like, in the end it is nothing else but applying "random" poisson weights to each event.

void	BoostMonitor (Int_t iTree)
	Fills the ROCIntegral vs Itree from the testSample for the monitoring plots during the training .

void	DeterminePreselectionCuts (const std::vector< const TMVA::Event * > &eventSample)
	Find useful preselection cuts that will be applied before and Decision Tree training.

void	GetBaggedSubSample (std::vector< const TMVA::Event * > &)
	Fills fEventSample with fBaggedSampleFraction*NEvents random training events.

Double_t	GetGradBoostMVA (const TMVA::Event *e, UInt_t nTrees)
	Returns MVA value: -1 for background, 1 for signal.

Double_t	GetMvaValue (Double_t err, Double_t errUpper, UInt_t useNTrees)
	Return the MVA value (range [-1;1]) that classifies the event according to the majority vote from the total number of decision trees.

Double_t	GradBoost (std::vector< const TMVA::Event * > &, DecisionTree *dt, UInt_t cls=0)
	Calculate the desired response value for each region.

Double_t	GradBoostRegression (std::vector< const TMVA::Event * > &, DecisionTree *dt)
	Implementation of M_TreeBoost using any loss function as described by Friedman 1999.

void	Init (void)
	Common initialisation with defaults for the BDT-Method.

void	InitGradBoost (std::vector< const TMVA::Event * > &)
	Initialize targets for first tree.

void	PreProcessNegativeEventWeights ()
	O.k.

Double_t	PrivateGetMvaValue (const TMVA::Event ev, Double_t err=nullptr, Double_t *errUpper=nullptr, UInt_t useNTrees=0)
	Return the MVA value (range [-1;1]) that classifies the event according to the majority vote from the total number of decision trees.

Double_t	RegBoost (std::vector< const TMVA::Event * > &, DecisionTree *dt)
	A special boosting only for Regression (not implemented).

void	UpdateTargets (std::vector< const TMVA::Event * > &, UInt_t cls=0)
	Calculate residual for all events.

void	UpdateTargetsRegression (std::vector< const TMVA::Event * > &, Bool_t first=kFALSE)
	Calculate residuals for all events and update targets for next iter.

Private Attributes
Double_t	fAdaBoostBeta
	beta parameter for AdaBoost algorithm

TString	fAdaBoostR2Loss
	loss type used in AdaBoostR2 (Linear,Quadratic or Exponential)

Bool_t	fAutomatic
	use user given prune strength or automatically determined one using a validation sample

Bool_t	fBaggedBoost
	turn bagging in combination with boost on/off

Bool_t	fBaggedGradBoost
	turn bagging in combination with grad boost on/off

Double_t	fBaggedSampleFraction
	relative size of bagged event sample to original sample size

TString	fBoostType
	string specifying the boost type

Double_t	fBoostWeight
	ntuple var: boost weight

std::vector< double >	fBoostWeights
	the weights applied in the individual boosts

Double_t	fCbb
	Cost factor.

Double_t	fCss
	Cost factor.

Double_t	fCtb_ss
	Cost factor.

Double_t	fCts_sb
	Cost factor.

Bool_t	fDoBoostMonitor
	create control plot with ROC integral vs tree number

Bool_t	fDoPreselection
	do or do not perform automatic pre-selection of 100% eff. cuts

Double_t	fErrorFraction
	ntuple var: misclassification error fraction

std::vector< const TMVA::Event * >	fEventSample
	the training events

std::vector< DecisionTree * >	fForest
	the collection of decision trees

Double_t	fFValidationEvents
	fraction of events to use for pruning

std::vector< Double_t >	fHighBkgCut

std::vector< Double_t >	fHighSigCut

Bool_t	fHistoricBool

Double_t	fHuberQuantile
	the option string determining the quantile for the Huber Loss Function in BDT regression.

Bool_t	fInverseBoostNegWeights
	boost ev. with neg. weights with 1/boostweight rather than boostweight

std::vector< Bool_t >	fIsHighBkgCut

std::vector< Bool_t >	fIsHighSigCut

std::vector< Bool_t >	fIsLowBkgCut

std::vector< Bool_t >	fIsLowSigCut

Int_t	fITree
	ntuple var: ith tree

std::map< const TMVA::Event *, LossFunctionEventInfo >	fLossFunctionEventInfo
	map event to true value, predicted value, and weight used by different loss functions for BDT regression

std::vector< Double_t >	fLowBkgCut

std::vector< Double_t >	fLowSigCut

UInt_t	fMaxDepth
	max depth

Double_t	fMinLinCorrForFisher
	the minimum linear correlation between two variables demanded for use in fisher criterium in node splitting

Int_t	fMinNodeEvents
	min number of events in node

Float_t	fMinNodeSize
	min percentage of training events in node

TString	fMinNodeSizeS
	string containing min percentage of training events in node

TTree *	fMonitorNtuple
	monitoring ntuple

Int_t	fNCuts
	grid used in cut applied in node splitting

TString	fNegWeightTreatment
	variable that holds the option of how to treat negative event weights in training

UInt_t	fNNodesMax
	max # of nodes

Double_t	fNodePurityLimit
	purity limit for sig/bkg nodes

Bool_t	fNoNegWeightsInTraining
	ignore negative event weights in the training

Int_t	fNTrees
	number of decision trees requested

Bool_t	fPairNegWeightsGlobal
	pair ev. with neg. and pos. weights in training sample and "annihilate" them

DecisionTree::EPruneMethod	fPruneMethod
	method used for pruning

TString	fPruneMethodS
	prune method option String

Double_t	fPruneStrength
	a parameter to set the "amount" of pruning..needs to be adjusted

Bool_t	fRandomisedTrees
	choose a random subset of possible cut variables at each node during training

LossFunctionBDT *	fRegressionLossFunctionBDTG

TString	fRegressionLossFunctionBDTGS
	the option string determining the loss function for BDT regression

std::map< const TMVA::Event *, std::vector< double > >	fResiduals
	individual event residuals for gradient boost

SeparationBase *	fSepType
	the separation used in node splitting

TString	fSepTypeS
	the separation (option string) used in node splitting

Double_t	fShrinkage
	learning rate for gradient boost;

Double_t	fSigToBkgFraction
	Signal to Background fraction assumed during training.

Bool_t	fSkipNormalization
	true for skipping normalization at initialization of trees

std::vector< const TMVA::Event * >	fSubSample
	subsample for bagged grad boost

std::vector< const TMVA::Event * > *	fTrainSample
	pointer to sample actually used in training (fEventSample or fSubSample) for example

Bool_t	fTrainWithNegWeights
	yes there are negative event weights and we don't ignore them

Bool_t	fUseExclusiveVars
	individual variables already used in fisher criterium are not anymore analysed individually for node splitting

Bool_t	fUseFisherCuts
	use multivariate splits using the Fisher criterium

UInt_t	fUseNTrainEvents
	number of randomly picked training events used in randomised (and bagged) trees

UInt_t	fUseNvars
	the number of variables used in the randomised tree splitting

Bool_t	fUsePoissonNvars
	use "fUseNvars" not as fixed number but as mean of a poisson distr. in each split

Bool_t	fUseYesNoLeaf
	use sig or bkg classification in leave nodes or sig/bkg

std::vector< const TMVA::Event * >	fValidationSample
	the Validation events

std::vector< Double_t >	fVariableImportance
	the relative importance of the different variables

Static Private Attributes
static const Int_t	fgDebugLevel = 0
	debug level determining some printout/control plots etc.

Additional Inherited Members
Public Types inherited from TMVA::MethodBase
enum	EWeightFileType { kROOT =0 , kTEXT }

Public Types inherited from TObject
enum	{ kIsOnHeap = 0x01000000 , kNotDeleted = 0x02000000 , kZombie = 0x04000000 , kInconsistent = 0x08000000 , kBitMask = 0x00ffffff }

enum	{ kSingleKey = (1ULL << ( 0 )) , kOverwrite = (1ULL << ( 1 )) , kWriteDelete = (1ULL << ( 2 )) }

enum	EDeprecatedStatusBits { kObjInCanvas = (1ULL << ( 3 )) }

enum	EStatusBits { kCanDelete = (1ULL << ( 0 )) , kMustCleanup = (1ULL << ( 3 )) , kIsReferenced = (1ULL << ( 4 )) , kHasUUID = (1ULL << ( 5 )) , kCannotPick = (1ULL << ( 6 )) , kNoContextMenu = (1ULL << ( 8 )) , kInvalidObject = (1ULL << ( 13 )) }

Public Attributes inherited from TMVA::MethodBase
Bool_t	fSetupCompleted

TrainingHistory	fTrainHistory

Protected Types inherited from TObject
enum	{ kOnlyPrepStep = (1ULL << ( 3 )) }

Protected Attributes inherited from TMVA::MethodBase
Types::EAnalysisType	fAnalysisType

UInt_t	fBackgroundClass

bool	fExitFromTraining = false

std::vector< TString > *	fInputVars

IPythonInteractive *	fInteractive = nullptr
	temporary dataset used when evaluating on a different data (used by MethodCategory::GetMvaValues)

UInt_t	fIPyCurrentIter = 0

UInt_t	fIPyMaxIter = 0

std::vector< Float_t > *	fMulticlassReturnVal

Int_t	fNbins

Int_t	fNbinsH

Int_t	fNbinsMVAoutput

Ranking *	fRanking

std::vector< Float_t > *	fRegressionReturnVal

Results *	fResults

UInt_t	fSignalClass

DataSet *	fTmpData = nullptr
	temporary event when testing on a different DataSet than the own one

const Event *	fTmpEvent

Protected Attributes inherited from TMVA::Configurable
MsgLogger *	fLogger
	! message logger

Protected Attributes inherited from TNamed
TString	fName

TString	fTitle

#include <TMVA/MethodBDT.h>

Inheritance diagram for TMVA::MethodBDT:

[legend]

Constructor & Destructor Documentation

◆ MethodBDT() [1/2]

TMVA::MethodBDT::MethodBDT	(	const TString &	jobName,
		const TString &	methodTitle,
		DataSetInfo &	theData,
		const TString &	theOption = `""`
	)

The standard constructor for the "boosted decision trees".

Definition at line 163 of file MethodBDT.cxx.

◆ MethodBDT() [2/2]

TMVA::MethodBDT::MethodBDT	(	DataSetInfo &	theData,
		const TString &	theWeightFile
	)

Definition at line 220 of file MethodBDT.cxx.

◆ ~MethodBDT()

TMVA::MethodBDT::~MethodBDT ( void )

virtual

Destructor.

Note: fEventSample and ValidationSample are already deleted at the end of TRAIN When they are not used anymore

Definition at line 753 of file MethodBDT.cxx.

Member Function Documentation

◆ AdaBoost()

Double_t TMVA::MethodBDT::AdaBoost	(	std::vector< const TMVA::Event * > &	eventSample,
		DecisionTree *	dt
	)

private

The AdaBoost implementation.

a new training sample is generated by weighting events that are misclassified by the decision tree. The weight applied is \( w = \frac{(1-err)}{err} \) or more general: \( w = (\frac{(1-err)}{err})^\beta \) where \(err\) is the fraction of misclassified events in the tree ( <0.5 assuming demanding the that previous selection was better than random guessing) and "beta" being a free parameter (standard: beta = 1) that modifies the boosting.

Definition at line 1845 of file MethodBDT.cxx.

◆ AdaBoostR2()

Double_t TMVA::MethodBDT::AdaBoostR2	(	std::vector< const TMVA::Event * > &	eventSample,
		DecisionTree *	dt
	)

private

Adaption of the AdaBoost to regression problems (see H.Drucker 1997).

Definition at line 2192 of file MethodBDT.cxx.

◆ AdaCost()

Double_t TMVA::MethodBDT::AdaCost	(	std::vector< const TMVA::Event * > &	eventSample,
		DecisionTree *	dt
	)

private

The AdaCost boosting algorithm takes a simple cost Matrix (currently fixed for all events... later could be modified to use individual cost matrices for each events as in the original paper...

              true_signal true_bkg
----------------------------------
sel_signal |   Css         Ctb_ss    Cxx.. in the range [0,1]
sel_bkg    |   Cts_sb      Cbb

and takes this into account when calculating the mis class. cost (former: error fraction):

err = sum_events ( weight* y_true*y_sel * beta(event)

Definition at line 2023 of file MethodBDT.cxx.

◆ AddWeightsXMLTo()

void TMVA::MethodBDT::AddWeightsXMLTo ( void * parent ) const

virtual

Write weights to XML.

Implements TMVA::MethodBase.

Definition at line 2309 of file MethodBDT.cxx.

◆ ApplyPreselectionCuts()

Double_t TMVA::MethodBDT::ApplyPreselectionCuts ( const Event * ev )

private

Apply the preselection cuts before even bothering about any Decision Trees in the GetMVA .

. --> -1 for background +1 for Signal

Definition at line 3132 of file MethodBDT.cxx.

◆ Bagging()

Double_t TMVA::MethodBDT::Bagging ( )

private

Call it boot-strapping, re-sampling or whatever you like, in the end it is nothing else but applying "random" poisson weights to each event.

Definition at line 2139 of file MethodBDT.cxx.

◆ Boost()

Double_t TMVA::MethodBDT::Boost	(	std::vector< const TMVA::Event * > &	eventSample,
		DecisionTree *	dt,
		UInt_t	cls = `0`
	)

Apply the boosting algorithm (the algorithm is selecte via the "option" given in the constructor.

The return value is the boosting weight.

Definition at line 1717 of file MethodBDT.cxx.

◆ BoostMonitor()

void TMVA::MethodBDT::BoostMonitor ( Int_t iTree )

private

Fills the ROCIntegral vs Itree from the testSample for the monitoring plots during the training .

. but using the testing events

Definition at line 1751 of file MethodBDT.cxx.

◆ Class()

static TClass * TMVA::MethodBDT::Class ( )

static

Returns: TClass describing this class

◆ Class_Name()

static const char * TMVA::MethodBDT::Class_Name ( )

static

Returns: Name of this class

◆ Class_Version()

static constexpr Version_t TMVA::MethodBDT::Class_Version ( )

inlinestaticconstexpr

Returns: Version of this class

Definition at line 305 of file MethodBDT.h.

◆ CreateRanking()

const TMVA::Ranking * TMVA::MethodBDT::CreateRanking ( )

virtual

Compute ranking of input variables.

Implements TMVA::MethodBase.

Definition at line 2682 of file MethodBDT.cxx.

◆ DeclareCompatibilityOptions()

void TMVA::MethodBDT::DeclareCompatibilityOptions ( )

protectedvirtual

Options that are used ONLY for the READER to ensure backward compatibility.

Reimplemented from TMVA::MethodBase.

Definition at line 454 of file MethodBDT.cxx.

◆ DeclareOptions()

void TMVA::MethodBDT::DeclareOptions ( )

virtual

Define the options (their key words).

That can be set in the option string.

know options:

nTrees number of trees in the forest to be created
BoostType the boosting type for the trees in the forest (AdaBoost e.t.c..). Known:
- AdaBoost
- AdaBoostR2 (Adaboost for regression)
- Bagging
- GradBoost
AdaBoostBeta the boosting parameter, beta, for AdaBoost
UseRandomisedTrees choose at each node splitting a random set of variables
UseNvars use UseNvars variables in randomised trees
UsePoisson Nvars use UseNvars not as fixed number but as mean of a poisson distribution
SeparationType the separation criterion applied in the node splitting. Known:
- GiniIndex
- MisClassificationError
- CrossEntropy
- SDivSqrtSPlusB
MinNodeSize: minimum percentage of training events in a leaf node (leaf criteria, stop splitting)
nCuts: the number of steps in the optimisation of the cut for a node (if < 0, then step size is determined by the events)
UseFisherCuts: use multivariate splits using the Fisher criterion
UseYesNoLeaf decide if the classification is done simply by the node type, or the S/B (from the training) in the leaf node
NodePurityLimit the minimum purity to classify a node as a signal node (used in pruning and boosting to determine misclassification error rate)
PruneMethod The Pruning method. Known:
- NoPruning // switch off pruning completely
- ExpectedError
- CostComplexity
PruneStrength a parameter to adjust the amount of pruning. Should be large enough such that overtraining is avoided.
PruningValFraction number of events to use for optimizing pruning (only if PruneStrength < 0, i.e. automatic pruning)
NegWeightTreatment
- IgnoreNegWeightsInTraining Ignore negative weight events in the training.
- DecreaseBoostWeight Boost ev. with neg. weight with 1/boostweight instead of boostweight
- PairNegWeightsGlobal Pair ev. with neg. and pos. weights in training sample and "annihilate" them
MaxDepth maximum depth of the decision tree allowed before further splitting is stopped
SkipNormalization Skip normalization at initialization, to keep expectation value of BDT output according to the fraction of events

Implements TMVA::MethodBase.

Definition at line 333 of file MethodBDT.cxx.

◆ DeclFileName()

static const char * TMVA::MethodBDT::DeclFileName ( )

inlinestatic

Returns: Name of the file containing the class declaration

Definition at line 305 of file MethodBDT.h.

◆ DeterminePreselectionCuts()

void TMVA::MethodBDT::DeterminePreselectionCuts ( const std::vector< const TMVA::Event * > & eventSample )

private

Find useful preselection cuts that will be applied before and Decision Tree training.

. (and of course also applied in the GetMVA .. --> -1 for background +1 for Signal)

Definition at line 3035 of file MethodBDT.cxx.

◆ GetBaggedSubSample()

void TMVA::MethodBDT::GetBaggedSubSample ( std::vector< const TMVA::Event * > & eventSample )

private

Fills fEventSample with fBaggedSampleFraction*NEvents random training events.

Definition at line 2150 of file MethodBDT.cxx.

◆ GetBoostWeights()

const std::vector< double > & TMVA::MethodBDT::GetBoostWeights ( ) const

inline

Definition at line 312 of file MethodBDT.h.

◆ GetForest()

const std::vector< TMVA::DecisionTree * > & TMVA::MethodBDT::GetForest ( ) const

inline

Definition at line 310 of file MethodBDT.h.

◆ GetGradBoostMVA()

Double_t TMVA::MethodBDT::GetGradBoostMVA	(	const TMVA::Event *	e,
		UInt_t	nTrees
	)

private

Returns MVA value: -1 for background, 1 for signal.

Definition at line 1420 of file MethodBDT.cxx.

◆ GetHelpMessage()

void TMVA::MethodBDT::GetHelpMessage ( ) const

virtual

Get help message text.

Implements TMVA::IMethod.

Definition at line 2699 of file MethodBDT.cxx.

◆ GetMulticlassValues()

const std::vector< Float_t > & TMVA::MethodBDT::GetMulticlassValues ( )

virtual

Get the multiclass MVA response for the BDT classifier.

Reimplemented from TMVA::MethodBase.

Definition at line 2494 of file MethodBDT.cxx.

◆ GetMvaValue() [1/2]

Double_t TMVA::MethodBDT::GetMvaValue	(	Double_t *	err,
		Double_t *	errUpper,
		UInt_t	useNTrees
	)

private

Return the MVA value (range [-1;1]) that classifies the event according to the majority vote from the total number of decision trees.

Definition at line 2451 of file MethodBDT.cxx.

◆ GetMvaValue() [2/2]

Double_t TMVA::MethodBDT::GetMvaValue	(	Double_t *	err = `nullptr`,
		Double_t *	errUpper = `nullptr`
	)

virtual

Implements TMVA::MethodBase.

Definition at line 2442 of file MethodBDT.cxx.

◆ GetNTrees()

UInt_t TMVA::MethodBDT::GetNTrees ( ) const

inline

Definition at line 112 of file MethodBDT.h.

◆ GetRegressionValues()

const std::vector< Float_t > & TMVA::MethodBDT::GetRegressionValues ( )

virtual

Get the regression value generated by the BDTs.

Reimplemented from TMVA::MethodBase.

Definition at line 2541 of file MethodBDT.cxx.

◆ GetTrainingEvents()

const std::vector< const TMVA::Event * > & TMVA::MethodBDT::GetTrainingEvents ( ) const

inline

Definition at line 311 of file MethodBDT.h.

◆ GetVariableImportance() [1/2]

vector< Double_t > TMVA::MethodBDT::GetVariableImportance ( )

Return the relative variable importance, normalized to all variables together having the importance 1.

The importance in evaluated as the total separation-gain that this variable had in the decision trees (weighted by the number of events)

Definition at line 2642 of file MethodBDT.cxx.

◆ GetVariableImportance() [2/2]

Double_t TMVA::MethodBDT::GetVariableImportance ( UInt_t ivar )

Returns the measure for the variable importance of variable "ivar" which is later used in GetVariableImportance() to calculate the relative variable importances.

Definition at line 2670 of file MethodBDT.cxx.

◆ GradBoost()

Double_t TMVA::MethodBDT::GradBoost	(	std::vector< const TMVA::Event * > &	eventSample,
		DecisionTree *	dt,
		UInt_t	cls = `0`
	)

private

Calculate the desired response value for each region.

Definition at line 1594 of file MethodBDT.cxx.

◆ GradBoostRegression()

Double_t TMVA::MethodBDT::GradBoostRegression	(	std::vector< const TMVA::Event * > &	eventSample,
		DecisionTree *	dt
	)

private

Implementation of M_TreeBoost using any loss function as described by Friedman 1999.

Definition at line 1628 of file MethodBDT.cxx.

◆ HasAnalysisType()

Bool_t TMVA::MethodBDT::HasAnalysisType	(	Types::EAnalysisType	type,
		UInt_t	numberClasses,
		UInt_t	numberTargets
	)

virtual

BDT can handle classification with multiple classes and regression with one regression-target.

Implements TMVA::IMethod.

Definition at line 280 of file MethodBDT.cxx.

◆ Init()

void TMVA::MethodBDT::Init ( void )

privatevirtual

Common initialisation with defaults for the BDT-Method.

Implements TMVA::MethodBase.

Definition at line 687 of file MethodBDT.cxx.

◆ InitEventSample()

void TMVA::MethodBDT::InitEventSample ( void )

Initialize the event sample (i.e. reset the boost-weights... etc).

Definition at line 761 of file MethodBDT.cxx.

◆ InitGradBoost()

void TMVA::MethodBDT::InitGradBoost ( std::vector< const TMVA::Event * > & eventSample )

private

Initialize targets for first tree.

Definition at line 1657 of file MethodBDT.cxx.

◆ IsA()

virtual TClass * TMVA::MethodBDT::IsA ( ) const

inlinevirtual

Returns: TClass describing current object

Reimplemented from TMVA::MethodBase.

Definition at line 305 of file MethodBDT.h.

◆ MakeClassInstantiateNode()

void TMVA::MethodBDT::MakeClassInstantiateNode	(	DecisionTreeNode *	n,
		std::ostream &	fout,
		const TString &	className
	)		const

Recursively descends a tree and writes the node instance to the output stream.

Definition at line 2990 of file MethodBDT.cxx.

◆ MakeClassSpecific()

void TMVA::MethodBDT::MakeClassSpecific	(	std::ostream &	fout,
		const TString &	className
	)		const

virtual

Make ROOT-independent C++ class for classifier response (classifier-specific implementation).

Reimplemented from TMVA::MethodBase.

Definition at line 2756 of file MethodBDT.cxx.

◆ MakeClassSpecificHeader()

void TMVA::MethodBDT::MakeClassSpecificHeader	(	std::ostream &	fout,
		const TString &	className
	)		const

virtual

Specific class header.

Reimplemented from TMVA::MethodBase.

Definition at line 2876 of file MethodBDT.cxx.

◆ OptimizeTuningParameters()

std::map< TString, Double_t > TMVA::MethodBDT::OptimizeTuningParameters	(	TString	fomType = `"ROCIntegral"`,
		TString	fitType = `"FitGA"`
	)

virtual

Call the Optimizer with the set of parameters and ranges that are meant to be tuned.

Reimplemented from TMVA::MethodBase.

Definition at line 1068 of file MethodBDT.cxx.

◆ PreProcessNegativeEventWeights()

void TMVA::MethodBDT::PreProcessNegativeEventWeights ( )

private

O.k.

you know there are events with negative event weights. This routine will remove them by pairing them with the closest event(s) of the same event class with positive weights A first attempt is "brute force", I dont' try to be clever using search trees etc, just quick and dirty to see if the result is any good

Definition at line 932 of file MethodBDT.cxx.

◆ PrivateGetMvaValue()

Double_t TMVA::MethodBDT::PrivateGetMvaValue	(	const TMVA::Event *	ev,
		Double_t *	err = `nullptr`,
		Double_t *	errUpper = `nullptr`,
		UInt_t	useNTrees = `0`
	)

private

Return the MVA value (range [-1;1]) that classifies the event according to the majority vote from the total number of decision trees.

Definition at line 2467 of file MethodBDT.cxx.

◆ ProcessOptions()

void TMVA::MethodBDT::ProcessOptions ( )

virtual

The option string is decoded, for available options see "DeclareOptions".

Implements TMVA::MethodBase.

Definition at line 470 of file MethodBDT.cxx.

◆ ReadWeightsFromStream() [1/3]

virtual void TMVA::MethodBase::ReadWeightsFromStream ( std::istream & )

virtual

Implements TMVA::MethodBase.

◆ ReadWeightsFromStream() [2/3]

void TMVA::MethodBDT::ReadWeightsFromStream ( std::istream & istr )

virtual

Read the weights (BDT coefficients).

Implements TMVA::MethodBase.

Definition at line 2407 of file MethodBDT.cxx.

◆ ReadWeightsFromStream() [3/3]

virtual void TMVA::MethodBase::ReadWeightsFromStream ( TFile & )

inlinevirtual

Reimplemented from TMVA::MethodBase.

Definition at line 266 of file MethodBase.h.

◆ ReadWeightsFromXML()

void TMVA::MethodBDT::ReadWeightsFromXML ( void * parent )

virtual

Reads the BDT from the xml file.

Implements TMVA::MethodBase.

Definition at line 2340 of file MethodBDT.cxx.

◆ RegBoost()

Double_t TMVA::MethodBDT::RegBoost	(	std::vector< const TMVA::Event * > &	,
		DecisionTree *	dt
	)

private

A special boosting only for Regression (not implemented).

Definition at line 2184 of file MethodBDT.cxx.

◆ Reset()

void TMVA::MethodBDT::Reset ( void )

virtual

Reset the method, as if it had just been instantiated (forget all training etc.).

Reimplemented from TMVA::MethodBase.

Definition at line 725 of file MethodBDT.cxx.

◆ SetAdaBoostBeta()

void TMVA::MethodBDT::SetAdaBoostBeta ( Double_t b )

inline

Definition at line 139 of file MethodBDT.h.

◆ SetBaggedSampleFraction()

void TMVA::MethodBDT::SetBaggedSampleFraction ( Double_t f )

inline

Definition at line 143 of file MethodBDT.h.

◆ SetMaxDepth()

void TMVA::MethodBDT::SetMaxDepth ( Int_t d )

inline

Definition at line 134 of file MethodBDT.h.

◆ SetMinNodeSize() [1/2]

void TMVA::MethodBDT::SetMinNodeSize ( Double_t sizeInPercent )

Definition at line 660 of file MethodBDT.cxx.

◆ SetMinNodeSize() [2/2]

void TMVA::MethodBDT::SetMinNodeSize ( TString sizeInPercent )

Definition at line 674 of file MethodBDT.cxx.

◆ SetNodePurityLimit()

void TMVA::MethodBDT::SetNodePurityLimit ( Double_t l )

inline

Definition at line 140 of file MethodBDT.h.

◆ SetNTrees()

void TMVA::MethodBDT::SetNTrees ( Int_t d )

inline

Definition at line 138 of file MethodBDT.h.

◆ SetShrinkage()

void TMVA::MethodBDT::SetShrinkage ( Double_t s )

inline

Definition at line 141 of file MethodBDT.h.

◆ SetTuneParameters()

void TMVA::MethodBDT::SetTuneParameters ( std::map< TString, Double_t > tuneParameters )

virtual

Set the tuning parameters according to the argument.

Reimplemented from TMVA::MethodBase.

Definition at line 1121 of file MethodBDT.cxx.

◆ SetUseNvars()

void TMVA::MethodBDT::SetUseNvars ( Int_t n )

inline

Definition at line 142 of file MethodBDT.h.

◆ Streamer()

virtual void TMVA::MethodBDT::Streamer ( TBuffer & )

virtual

Reimplemented from TMVA::MethodBase.

◆ StreamerNVirtual()

void TMVA::MethodBDT::StreamerNVirtual ( TBuffer & ClassDef_StreamerNVirtual_b )

inline

Definition at line 305 of file MethodBDT.h.

◆ TestTreeQuality()

Double_t TMVA::MethodBDT::TestTreeQuality ( DecisionTree * dt )

Test the tree quality.. in terms of Misclassification.

Definition at line 1696 of file MethodBDT.cxx.

◆ Train()

void TMVA::MethodBDT::Train ( void )

virtual

BDT training.

Implements TMVA::MethodBase.

Definition at line 1142 of file MethodBDT.cxx.

◆ UpdateTargets()

void TMVA::MethodBDT::UpdateTargets	(	std::vector< const TMVA::Event * > &	eventSample,
		UInt_t	cls = `0`
	)

private

Calculate residual for all events.

Definition at line 1434 of file MethodBDT.cxx.

◆ UpdateTargetsRegression()

void TMVA::MethodBDT::UpdateTargetsRegression	(	std::vector< const TMVA::Event * > &	eventSample,
		Bool_t	first = `kFALSE`
	)

private

Calculate residuals for all events and update targets for next iter.

Parameters

[in]	eventSample	The collection of events currently under training.
[in]	first	Should be true when called before the first boosting iteration has been run

Definition at line 1556 of file MethodBDT.cxx.

◆ WriteMonitoringHistosToFile()

void TMVA::MethodBDT::WriteMonitoringHistosToFile ( void ) const

virtual

Here we could write some histograms created during the processing to the output file.

Reimplemented from TMVA::MethodBase.

Definition at line 2627 of file MethodBDT.cxx.

Member Data Documentation

◆ fAdaBoostBeta

Double_t TMVA::MethodBDT::fAdaBoostBeta

private

beta parameter for AdaBoost algorithm

Definition at line 216 of file MethodBDT.h.

◆ fAdaBoostR2Loss

TString TMVA::MethodBDT::fAdaBoostR2Loss

private

loss type used in AdaBoostR2 (Linear,Quadratic or Exponential)

Definition at line 217 of file MethodBDT.h.

◆ fAutomatic

Bool_t TMVA::MethodBDT::fAutomatic

private

use user given prune strength or automatically determined one using a validation sample

Definition at line 248 of file MethodBDT.h.

◆ fBaggedBoost

Bool_t TMVA::MethodBDT::fBaggedBoost

private

turn bagging in combination with boost on/off

Definition at line 220 of file MethodBDT.h.

◆ fBaggedGradBoost

Bool_t TMVA::MethodBDT::fBaggedGradBoost

private

turn bagging in combination with grad boost on/off

Definition at line 221 of file MethodBDT.h.

◆ fBaggedSampleFraction

Double_t TMVA::MethodBDT::fBaggedSampleFraction

private

relative size of bagged event sample to original sample size

Definition at line 254 of file MethodBDT.h.

◆ fBoostType

TString TMVA::MethodBDT::fBoostType

private

string specifying the boost type

Definition at line 215 of file MethodBDT.h.

◆ fBoostWeight

Double_t TMVA::MethodBDT::fBoostWeight

private

ntuple var: boost weight

Definition at line 266 of file MethodBDT.h.

◆ fBoostWeights

std::vector<double> TMVA::MethodBDT::fBoostWeights

private

the weights applied in the individual boosts

Definition at line 213 of file MethodBDT.h.

◆ fCbb

Double_t TMVA::MethodBDT::fCbb

private

Cost factor.

Definition at line 272 of file MethodBDT.h.

◆ fCss

Double_t TMVA::MethodBDT::fCss

private

Cost factor.

Definition at line 269 of file MethodBDT.h.

◆ fCtb_ss

Double_t TMVA::MethodBDT::fCtb_ss

private

Cost factor.

Definition at line 271 of file MethodBDT.h.

◆ fCts_sb

Double_t TMVA::MethodBDT::fCts_sb

private

Cost factor.

Definition at line 270 of file MethodBDT.h.

◆ fDoBoostMonitor

Bool_t TMVA::MethodBDT::fDoBoostMonitor

private

create control plot with ROC integral vs tree number

Definition at line 260 of file MethodBDT.h.

◆ fDoPreselection

Bool_t TMVA::MethodBDT::fDoPreselection

private

do or do not perform automatic pre-selection of 100% eff. cuts

Definition at line 274 of file MethodBDT.h.

◆ fErrorFraction

Double_t TMVA::MethodBDT::fErrorFraction

private

ntuple var: misclassification error fraction

Definition at line 267 of file MethodBDT.h.

◆ fEventSample

std::vector<const TMVA::Event*> TMVA::MethodBDT::fEventSample

private

the training events

Definition at line 206 of file MethodBDT.h.

◆ fForest

std::vector<DecisionTree*> TMVA::MethodBDT::fForest

private

the collection of decision trees

Definition at line 212 of file MethodBDT.h.

◆ fFValidationEvents

Double_t TMVA::MethodBDT::fFValidationEvents

private

fraction of events to use for pruning

Definition at line 247 of file MethodBDT.h.

◆ fgDebugLevel

const Int_t TMVA::MethodBDT::fgDebugLevel = 0

staticprivate

debug level determining some printout/control plots etc.

Definition at line 302 of file MethodBDT.h.

◆ fHighBkgCut

std::vector<Double_t> TMVA::MethodBDT::fHighBkgCut

private

Definition at line 287 of file MethodBDT.h.

◆ fHighSigCut

std::vector<Double_t> TMVA::MethodBDT::fHighSigCut

private

Definition at line 286 of file MethodBDT.h.

◆ fHistoricBool

Bool_t TMVA::MethodBDT::fHistoricBool

private

Definition at line 294 of file MethodBDT.h.

◆ fHuberQuantile

Double_t TMVA::MethodBDT::fHuberQuantile

private

the option string determining the quantile for the Huber Loss Function in BDT regression.

Definition at line 297 of file MethodBDT.h.

◆ fInverseBoostNegWeights

Bool_t TMVA::MethodBDT::fInverseBoostNegWeights

private

boost ev. with neg. weights with 1/boostweight rather than boostweight

Definition at line 257 of file MethodBDT.h.

◆ fIsHighBkgCut

std::vector<Bool_t> TMVA::MethodBDT::fIsHighBkgCut

private

Definition at line 292 of file MethodBDT.h.

◆ fIsHighSigCut

std::vector<Bool_t> TMVA::MethodBDT::fIsHighSigCut

private

Definition at line 291 of file MethodBDT.h.

◆ fIsLowBkgCut

std::vector<Bool_t> TMVA::MethodBDT::fIsLowBkgCut

private

Definition at line 290 of file MethodBDT.h.

◆ fIsLowSigCut

std::vector<Bool_t> TMVA::MethodBDT::fIsLowSigCut

private

Definition at line 289 of file MethodBDT.h.

◆ fITree

Int_t TMVA::MethodBDT::fITree

private

ntuple var: ith tree

Definition at line 265 of file MethodBDT.h.

◆ fLossFunctionEventInfo

std::map< const TMVA::Event*, LossFunctionEventInfo> TMVA::MethodBDT::fLossFunctionEventInfo

private

map event to true value, predicted value, and weight used by different loss functions for BDT regression

Definition at line 224 of file MethodBDT.h.

◆ fLowBkgCut

std::vector<Double_t> TMVA::MethodBDT::fLowBkgCut

private

Definition at line 285 of file MethodBDT.h.

◆ fLowSigCut

std::vector<Double_t> TMVA::MethodBDT::fLowSigCut

private

Definition at line 284 of file MethodBDT.h.

◆ fMaxDepth

UInt_t TMVA::MethodBDT::fMaxDepth

private

max depth

Definition at line 242 of file MethodBDT.h.

◆ fMinLinCorrForFisher

Double_t TMVA::MethodBDT::fMinLinCorrForFisher

private

the minimum linear correlation between two variables demanded for use in fisher criterium in node splitting

Definition at line 237 of file MethodBDT.h.

◆ fMinNodeEvents

Int_t TMVA::MethodBDT::fMinNodeEvents

private

min number of events in node

Definition at line 231 of file MethodBDT.h.

◆ fMinNodeSize

Float_t TMVA::MethodBDT::fMinNodeSize

private

min percentage of training events in node

Definition at line 232 of file MethodBDT.h.

◆ fMinNodeSizeS

TString TMVA::MethodBDT::fMinNodeSizeS

private

string containing min percentage of training events in node

Definition at line 233 of file MethodBDT.h.

◆ fMonitorNtuple

TTree* TMVA::MethodBDT::fMonitorNtuple

private

monitoring ntuple

Definition at line 264 of file MethodBDT.h.

◆ fNCuts

Int_t TMVA::MethodBDT::fNCuts

private

grid used in cut applied in node splitting

Definition at line 235 of file MethodBDT.h.

◆ fNegWeightTreatment

TString TMVA::MethodBDT::fNegWeightTreatment

private

variable that holds the option of how to treat negative event weights in training

Definition at line 255 of file MethodBDT.h.

◆ fNNodesMax

UInt_t TMVA::MethodBDT::fNNodesMax

private

max # of nodes

Definition at line 241 of file MethodBDT.h.

◆ fNodePurityLimit

Double_t TMVA::MethodBDT::fNodePurityLimit

private

purity limit for sig/bkg nodes

Definition at line 240 of file MethodBDT.h.

◆ fNoNegWeightsInTraining

Bool_t TMVA::MethodBDT::fNoNegWeightsInTraining

private

ignore negative event weights in the training

Definition at line 256 of file MethodBDT.h.

◆ fNTrees

Int_t TMVA::MethodBDT::fNTrees

private

number of decision trees requested

Definition at line 211 of file MethodBDT.h.

◆ fPairNegWeightsGlobal

Bool_t TMVA::MethodBDT::fPairNegWeightsGlobal

private

pair ev. with neg. and pos. weights in training sample and "annihilate" them

Definition at line 258 of file MethodBDT.h.

◆ fPruneMethod

DecisionTree::EPruneMethod TMVA::MethodBDT::fPruneMethod

private

method used for pruning

Definition at line 244 of file MethodBDT.h.

◆ fPruneMethodS

TString TMVA::MethodBDT::fPruneMethodS

private

prune method option String

Definition at line 245 of file MethodBDT.h.

◆ fPruneStrength

Double_t TMVA::MethodBDT::fPruneStrength

private

a parameter to set the "amount" of pruning..needs to be adjusted

Definition at line 246 of file MethodBDT.h.

◆ fRandomisedTrees

Bool_t TMVA::MethodBDT::fRandomisedTrees

private

choose a random subset of possible cut variables at each node during training

Definition at line 249 of file MethodBDT.h.

◆ fRegressionLossFunctionBDTG

LossFunctionBDT* TMVA::MethodBDT::fRegressionLossFunctionBDTG

private

Definition at line 299 of file MethodBDT.h.

◆ fRegressionLossFunctionBDTGS

TString TMVA::MethodBDT::fRegressionLossFunctionBDTGS

private

the option string determining the loss function for BDT regression

Definition at line 296 of file MethodBDT.h.

◆ fResiduals

std::map< const TMVA::Event*,std::vector<double> > TMVA::MethodBDT::fResiduals

private

individual event residuals for gradient boost

Definition at line 226 of file MethodBDT.h.

◆ fSepType

SeparationBase* TMVA::MethodBDT::fSepType

private

the separation used in node splitting

Definition at line 229 of file MethodBDT.h.

◆ fSepTypeS

TString TMVA::MethodBDT::fSepTypeS

private

the separation (option string) used in node splitting

Definition at line 230 of file MethodBDT.h.

◆ fShrinkage

Double_t TMVA::MethodBDT::fShrinkage

private

learning rate for gradient boost;

Definition at line 219 of file MethodBDT.h.

◆ fSigToBkgFraction

Double_t TMVA::MethodBDT::fSigToBkgFraction

private

Signal to Background fraction assumed during training.

Definition at line 214 of file MethodBDT.h.

◆ fSkipNormalization

Bool_t TMVA::MethodBDT::fSkipNormalization

private

true for skipping normalization at initialization of trees

Definition at line 276 of file MethodBDT.h.

◆ fSubSample

std::vector<const TMVA::Event*> TMVA::MethodBDT::fSubSample

private

subsample for bagged grad boost

Definition at line 208 of file MethodBDT.h.

◆ fTrainSample

std::vector<const TMVA::Event*>* TMVA::MethodBDT::fTrainSample

private

pointer to sample actually used in training (fEventSample or fSubSample) for example

Definition at line 209 of file MethodBDT.h.

◆ fTrainWithNegWeights

Bool_t TMVA::MethodBDT::fTrainWithNegWeights

private

yes there are negative event weights and we don't ignore them

Definition at line 259 of file MethodBDT.h.

◆ fUseExclusiveVars

Bool_t TMVA::MethodBDT::fUseExclusiveVars

private

individual variables already used in fisher criterium are not anymore analysed individually for node splitting

Definition at line 238 of file MethodBDT.h.

◆ fUseFisherCuts

Bool_t TMVA::MethodBDT::fUseFisherCuts

private

use multivariate splits using the Fisher criterium

Definition at line 236 of file MethodBDT.h.

◆ fUseNTrainEvents

UInt_t TMVA::MethodBDT::fUseNTrainEvents

private

number of randomly picked training events used in randomised (and bagged) trees

Definition at line 252 of file MethodBDT.h.

◆ fUseNvars

UInt_t TMVA::MethodBDT::fUseNvars

private

the number of variables used in the randomised tree splitting

Definition at line 250 of file MethodBDT.h.

◆ fUsePoissonNvars

Bool_t TMVA::MethodBDT::fUsePoissonNvars

private

use "fUseNvars" not as fixed number but as mean of a poisson distr. in each split

Definition at line 251 of file MethodBDT.h.

◆ fUseYesNoLeaf

Bool_t TMVA::MethodBDT::fUseYesNoLeaf

private

use sig or bkg classification in leave nodes or sig/bkg

Definition at line 239 of file MethodBDT.h.

◆ fValidationSample

std::vector<const TMVA::Event*> TMVA::MethodBDT::fValidationSample

private

the Validation events

Definition at line 207 of file MethodBDT.h.

◆ fVariableImportance

std::vector<Double_t> TMVA::MethodBDT::fVariableImportance

private

the relative importance of the different variables

Definition at line 278 of file MethodBDT.h.

Libraries for TMVA::MethodBDT:

[legend]

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

tmva/tmva/inc/TMVA/MethodBDT.h
tmva/tmva/src/MethodBDT.cxx

Decision trees:

Boosting:

Bagging:

Random Trees:

Analysis:

Public Member Functions

Static Public Member Functions

Protected Member Functions

Private Member Functions

Private Attributes

Static Private Attributes

Additional Inherited Members

Constructor & Destructor Documentation

◆ MethodBDT() [1/2]

◆ MethodBDT() [2/2]

◆ ~MethodBDT()

Member Function Documentation

◆ AdaBoost()

◆ AdaBoostR2()

◆ AdaCost()

◆ AddWeightsXMLTo()

◆ ApplyPreselectionCuts()

◆ Bagging()

◆ Boost()

◆ BoostMonitor()

◆ Class()

◆ Class_Name()

◆ Class_Version()

◆ CreateRanking()

◆ DeclareCompatibilityOptions()

◆ DeclareOptions()

◆ DeclFileName()

◆ DeterminePreselectionCuts()

◆ GetBaggedSubSample()

◆ GetBoostWeights()

◆ GetForest()

◆ GetGradBoostMVA()

◆ GetHelpMessage()

◆ GetMulticlassValues()

◆ GetMvaValue() [1/2]

◆ GetMvaValue() [2/2]

◆ GetNTrees()

◆ GetRegressionValues()

◆ GetTrainingEvents()

◆ GetVariableImportance() [1/2]

◆ GetVariableImportance() [2/2]

◆ GradBoost()

◆ GradBoostRegression()

◆ HasAnalysisType()

◆ Init()

◆ InitEventSample()

◆ InitGradBoost()

◆ IsA()

◆ MakeClassInstantiateNode()

◆ MakeClassSpecific()

◆ MakeClassSpecificHeader()

◆ OptimizeTuningParameters()

◆ PreProcessNegativeEventWeights()

◆ PrivateGetMvaValue()

◆ ProcessOptions()

◆ ReadWeightsFromStream() [1/3]

◆ ReadWeightsFromStream() [2/3]

◆ ReadWeightsFromStream() [3/3]

◆ ReadWeightsFromXML()

◆ RegBoost()

◆ Reset()

◆ SetAdaBoostBeta()

◆ SetBaggedSampleFraction()

◆ SetMaxDepth()

◆ SetMinNodeSize() [1/2]

◆ SetMinNodeSize() [2/2]

◆ SetNodePurityLimit()

◆ SetNTrees()

◆ SetShrinkage()

◆ SetTuneParameters()

◆ SetUseNvars()

◆ Streamer()

◆ StreamerNVirtual()

◆ TestTreeQuality()

◆ Train()