xgboost::TreeUpdater Class Reference

interface of tree update module, that performs update of a tree. More...

#include <tree_updater.h>

Inheritance diagram for xgboost::TreeUpdater:

Public Member Functions
	TreeUpdater (const Context *ctx)
	~TreeUpdater () override=default
	virtual destructor
virtual void	Configure (const Args &args)=0
	Initialize the updater with given arguments.
virtual bool	CanModifyTree () const
	Whether this updater can be used for updating existing trees.
virtual bool	HasNodePosition () const
	Wether the out_position in Update is valid. This determines whether adaptive tree can be used.
virtual void	Update (tree::TrainParam const *param, HostDeviceVector< GradientPair > *gpair, DMatrix *data, common::Span< HostDeviceVector< bst_node_t > > out_position, const std::vector< RegTree * > &out_trees)=0
	perform update to the tree models
virtual bool	UpdatePredictionCache (const DMatrix *, linalg::MatrixView< float >)
	determines whether updater has enough knowledge about a given dataset to quickly update prediction cache its training data and performs the update if possible.
virtual char const *	Name () const =0
Public Member Functions inherited from xgboost::Configurable
virtual void	LoadConfig (Json const &in)=0
	Load configuration from JSON object.
virtual void	SaveConfig (Json *out) const =0
	Save configuration to JSON object.

Static Public Member Functions
static TreeUpdater *	Create (const std::string &name, Context const *ctx, ObjInfo const *task)
	Create a tree updater given name.

Protected Attributes
Context const *	ctx_ = nullptr

Detailed Description

interface of tree update module, that performs update of a tree.

Member Function Documentation

CanModifyTree()

virtual bool xgboost::TreeUpdater::CanModifyTree ( ) const

inlinevirtual

Whether this updater can be used for updating existing trees.

Some updaters are used for building new trees (like hist), while some others are used for modifying existing trees (like prune). Return true if it can modify existing trees.

Reimplemented in xgboost::tree::TreePruner, and xgboost::tree::TreeRefresher.

Configure()

virtual void xgboost::TreeUpdater::Configure ( const Args &  args )

pure virtual

Initialize the updater with given arguments.

Parameters

args	arguments to the objective function.

Implemented in xgboost::tree::GlobalApproxUpdater, xgboost::tree::QuantileHistMaker, xgboost::tree::TreeRefresher, xgboost::tree::TreeSyncher, xgboost::tree::ColMaker, and xgboost::tree::TreePruner.

Create()

TreeUpdater * xgboost::TreeUpdater::Create ( const std::string &  name, Context const *  ctx, ObjInfo const *  task  )

static

Create a tree updater given name.

Parameters

name	Name of the tree updater.
ctx	A global runtime parameter
task	Infomation about the objective.

HasNodePosition()

virtual bool xgboost::TreeUpdater::HasNodePosition ( ) const

inlinevirtual

Wether the out_position in Update is valid. This determines whether adaptive tree can be used.

Reimplemented in xgboost::tree::GlobalApproxUpdater, and xgboost::tree::QuantileHistMaker.

Update()

virtual void xgboost::TreeUpdater::Update ( tree::TrainParam const *  param, HostDeviceVector< GradientPair > *  gpair, DMatrix *  data, common::Span< HostDeviceVector< bst_node_t > >  out_position, const std::vector< RegTree * > &  out_trees  )

pure virtual

perform update to the tree models

Parameters

param	Hyper-parameter for constructing trees.
gpair	the gradient pair statistics of the data
data	The data matrix passed to the updater.
out_position	The leaf index for each row. The index is negated if that row is removed during sampling. So the 3th node is ~3.
out_trees	references the trees to be updated, updater will change the content of trees note: all the trees in the vector are updated, with the same statistics, but maybe different random seeds, usually one tree is passed in at a time, there can be multiple trees when we train random forest style model

Implemented in xgboost::tree::TreeSyncher, xgboost::tree::ColMaker, xgboost::tree::GlobalApproxUpdater, xgboost::tree::TreePruner, xgboost::tree::QuantileHistMaker, and xgboost::tree::TreeRefresher.

UpdatePredictionCache()

virtual bool xgboost::TreeUpdater::UpdatePredictionCache ( const DMatrix *  , linalg::MatrixView< float >    )

inlinevirtual

determines whether updater has enough knowledge about a given dataset to quickly update prediction cache its training data and performs the update if possible.

Parameters

data	data matrix
out_preds	prediction cache to be updated

Returns

boolean indicating whether updater has capability to update the prediction cache. If true, the prediction cache will have been updated by the time this function returns.

Reimplemented in xgboost::tree::GlobalApproxUpdater, and xgboost::tree::QuantileHistMaker.

---

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

• External/xgboost/include/xgboost/tree_updater.h
• External/xgboost/src/tree/tree_updater.cc