gradient boosted linear model More...

Inheritance diagram for xgboost::gbm::GBLinear:

Public Member Functions
	GBLinear (LearnerModelParam const learner_model_param, Context const ctx)

void	Configure (const Args &cfg) override
	Set the configuration of gradient boosting. User must call configure once before InitModel and Training.

int32_t	BoostedRounds () const override
	Return number of boosted rounds.

bool	ModelFitted () const override
	Whether the model has already been trained.

void	Load (dmlc::Stream *fi) override
	load model from stream

void	Save (dmlc::Stream *fo) const override
	save model to stream.

void	SaveModel (Json *p_out) const override
	saves the model config to a JSON object

void	LoadModel (Json const &in) override
	load the model from a JSON object

void	LoadConfig (Json const &in) override
	Load configuration from JSON object.

void	SaveConfig (Json *p_out) const override
	Save configuration to JSON object.

void	DoBoost (DMatrix p_fmat, HostDeviceVector< GradientPair > in_gpair, PredictionCacheEntry , ObjFunction const ) override
	perform update to the model(boosting)

void	PredictBatch (DMatrix p_fmat, PredictionCacheEntry predts, bool, bst_layer_t layer_begin, bst_layer_t) override
	Generate predictions for given feature matrix.

void	PredictInstance (const SparsePage::Inst &inst, std::vector< bst_float > *out_preds, uint32_t layer_begin, uint32_t) override

void	PredictLeaf (DMatrix , HostDeviceVector< bst_float > , unsigned, unsigned) override
	predict the leaf index of each tree, the output will be nsample * ntree vector this is only valid in gbtree predictor

void	PredictContribution (DMatrix p_fmat, HostDeviceVector< bst_float > out_contribs, bst_layer_t layer_begin, bst_layer_t, bool) override
	feature contributions to individual predictions; the output will be a vector of length (nfeats + 1) * num_output_group * nsample, arranged in that order

void	PredictInteractionContributions (DMatrix p_fmat, HostDeviceVector< float > out_contribs, bst_layer_t layer_begin, bst_layer_t, bool) override

std::vector< std::string >	DumpModel (const FeatureMap &fmap, bool with_stats, std::string format) const override
	dump the model in the requested format

void	FeatureScore (std::string const &importance_type, common::Span< int32_t const > trees, std::vector< bst_feature_t > out_features, std::vector< float > out_scores) const override

bool	UseGPU () const override
	Whether the current booster uses GPU.

Public Member Functions inherited from xgboost::GradientBooster
	~GradientBooster () override=default
	virtual destructor

virtual void	Slice (bst_layer_t, bst_layer_t, bst_layer_t, GradientBooster , bool ) const
	Slice a model using boosting index.

virtual void	InplacePredict (std::shared_ptr< DMatrix >, float, PredictionCacheEntry *, bst_layer_t, bst_layer_t) const
	Inplace prediction.

virtual void	PredictInstance (const SparsePage::Inst &inst, std::vector< bst_float > *out_preds, unsigned layer_begin, unsigned layer_end)=0
	online prediction function, predict score for one instance at a time NOTE: use the batch prediction interface if possible, batch prediction is usually more efficient than online prediction This function is NOT threadsafe, make sure you only call from one thread

Protected Member Functions
void	PredictBatchInternal (DMatrix p_fmat, std::vector< bst_float > out_preds)

bool	CheckConvergence ()

void	LazySumWeights (DMatrix *p_fmat)

void	Pred (const SparsePage::Inst &inst, bst_float *preds, int gid, bst_float base)

Protected Member Functions inherited from xgboost::GradientBooster
	GradientBooster (Context const *ctx)

Protected Attributes
LearnerModelParam const *	learner_model_param_

GBLinearModel	model_

GBLinearModel	previous_model_

GBLinearTrainParam	param_

std::unique_ptr< LinearUpdater >	updater_

double	sum_instance_weight_ {}

bool	sum_weight_complete_ {false}

common::Monitor	monitor_

bool	is_converged_ {false}

Protected Attributes inherited from xgboost::GradientBooster
Context const *	ctx_

Additional Inherited Members
Static Public Member Functions inherited from xgboost::GradientBooster
static GradientBooster *	Create (const std::string &name, Context const ctx, LearnerModelParam const learner_model_param)
	create a gradient booster from given name

Detailed Description

gradient boosted linear model

Member Function Documentation

◆ BoostedRounds()

int32_t xgboost::gbm::GBLinear::BoostedRounds ( ) const

inlineoverridevirtual

Return number of boosted rounds.

Implements xgboost::GradientBooster.

◆ Configure()

void xgboost::gbm::GBLinear::Configure ( const Args & cfg )

inlineoverridevirtual

Set the configuration of gradient boosting. User must call configure once before InitModel and Training.

Parameters

cfg	configurations on both training and model parameters.

Implements xgboost::GradientBooster.

◆ DoBoost()

void xgboost::gbm::GBLinear::DoBoost	(	DMatrix *	p_fmat,
		HostDeviceVector< GradientPair > *	in_gpair,
		PredictionCacheEntry *	,
		ObjFunction const *	obj
	)

inlineoverridevirtual

perform update to the model(boosting)

Parameters

p_fmat	feature matrix that provide access to features
in_gpair	address of the gradient pair statistics of the data
prediction	The output prediction cache entry that needs to be updated. the booster may change content of gpair

Implements xgboost::GradientBooster.

◆ DumpModel()

std::vector< std::string > xgboost::gbm::GBLinear::DumpModel	(	const FeatureMap &	fmap,
		bool	with_stats,
		std::string	format
	)		const

inlineoverridevirtual

dump the model in the requested format

Parameters

fmap	feature map that may help give interpretations of feature
with_stats	extra statistics while dumping model
format	the format to dump the model in

Returns: a vector of dump for boosters.

Implements xgboost::GradientBooster.

◆ FeatureScore()

void xgboost::gbm::GBLinear::FeatureScore	(	std::string const &	importance_type,
		common::Span< int32_t const >	trees,
		std::vector< bst_feature_t > *	out_features,
		std::vector< float > *	out_scores
	)		const

inlineoverridevirtual

Implements xgboost::GradientBooster.

◆ Load()

void xgboost::gbm::GBLinear::Load ( dmlc::Stream * fi )

inlineoverridevirtual

load model from stream

Parameters

fi	input stream.

Implements xgboost::GradientBooster.

◆ LoadConfig()

void xgboost::gbm::GBLinear::LoadConfig ( Json const & in )

inlineoverridevirtual

Load configuration from JSON object.

Parameters

in	JSON object containing the configuration

Implements xgboost::Configurable.

◆ LoadModel()

void xgboost::gbm::GBLinear::LoadModel ( Json const & in )

inlineoverridevirtual

load the model from a JSON object

Parameters

in	JSON object where to load the model from

Implements xgboost::Model.

◆ ModelFitted()

bool xgboost::gbm::GBLinear::ModelFitted ( ) const

inlineoverridevirtual

Whether the model has already been trained.

When tree booster is chosen, then returns true when there are existing trees.

Implements xgboost::GradientBooster.

◆ PredictBatch()

void xgboost::gbm::GBLinear::PredictBatch	(	DMatrix *	dmat,
		PredictionCacheEntry *	out_preds,
		bool	training,
		bst_layer_t	begin,
		bst_layer_t	end
	)

inlineoverridevirtual

Generate predictions for given feature matrix.

Parameters

dmat	The feature matrix.
out_preds	output vector to hold the predictions
training	Whether the prediction value is used for training. For dart booster drop out is performed during training.
begin	Beginning of boosted tree layer used for prediction.
end	End of booster layer. 0 means do not limit trees.

Implements xgboost::GradientBooster.

◆ PredictContribution()

void xgboost::gbm::GBLinear::PredictContribution	(	DMatrix *	dmat,
		HostDeviceVector< bst_float > *	out_contribs,
		bst_layer_t	layer_begin,
		bst_layer_t	layer_end,
		bool	approximate
	)

inlineoverridevirtual

feature contributions to individual predictions; the output will be a vector of length (nfeats + 1) * num_output_group * nsample, arranged in that order

Parameters

dmat	feature matrix
out_contribs	output vector to hold the contributions
layer_begin	Beginning of boosted tree layer used for prediction.
layer_end	End of booster layer. 0 means do not limit trees.
approximate	use a faster (inconsistent) approximation of SHAP values

Implements xgboost::GradientBooster.

◆ PredictInteractionContributions()

void xgboost::gbm::GBLinear::PredictInteractionContributions	(	DMatrix *	p_fmat,
		HostDeviceVector< float > *	out_contribs,
		bst_layer_t	layer_begin,
		bst_layer_t	,
		bool
	)

inlineoverridevirtual

Implements xgboost::GradientBooster.

◆ PredictLeaf()

void xgboost::gbm::GBLinear::PredictLeaf	(	DMatrix *	dmat,
		HostDeviceVector< bst_float > *	out_preds,
		unsigned	layer_begin,
		unsigned	layer_end
	)

inlineoverridevirtual

predict the leaf index of each tree, the output will be nsample * ntree vector this is only valid in gbtree predictor

Parameters

dmat	feature matrix
out_preds	output vector to hold the predictions
layer_begin	Beginning of boosted tree layer used for prediction.
layer_end	End of booster layer. 0 means do not limit trees.

Implements xgboost::GradientBooster.

◆ Save()

void xgboost::gbm::GBLinear::Save ( dmlc::Stream * fo ) const

inlineoverridevirtual

save model to stream.

Parameters

fo	output stream

Implements xgboost::GradientBooster.

◆ SaveConfig()

void xgboost::gbm::GBLinear::SaveConfig ( Json * out ) const

inlineoverridevirtual

Save configuration to JSON object.

Parameters

out	pointer to output JSON object

Implements xgboost::Configurable.

◆ SaveModel()

void xgboost::gbm::GBLinear::SaveModel ( Json * out ) const

inlineoverridevirtual

saves the model config to a JSON object

Parameters

out	JSON container where to save the model to

Implements xgboost::Model.

◆ UseGPU()

bool xgboost::gbm::GBLinear::UseGPU ( ) const

inlineoverridevirtual

Whether the current booster uses GPU.

Implements xgboost::GradientBooster.

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

External/xgboost/src/gbm/gblinear.cc

Public Member Functions

Protected Member Functions

Protected Attributes

Additional Inherited Members

Detailed Description

Member Function Documentation

◆ BoostedRounds()

◆ Configure()

◆ DoBoost()

◆ DumpModel()

◆ FeatureScore()

◆ Load()

◆ LoadConfig()

◆ LoadModel()

◆ ModelFitted()

◆ PredictBatch()

◆ PredictContribution()

◆ PredictInteractionContributions()

◆ PredictLeaf()

◆ Save()

◆ SaveConfig()

◆ SaveModel()

◆ UseGPU()