Model builder interface. More...

#include <model_builder.h>

Collaboration diagram for treelite::model_builder::ModelBuilder:

Public Member Functions
virtual void	StartTree ()=0
	Start a new tree. More...

virtual void	EndTree ()=0
	End the current tree. More...

virtual void	StartNode (int node_key)=0
	Start a new node. More...

virtual void	EndNode ()=0
	End the current node. More...

virtual void	NumericalTest (std::int32_t split_index, double threshold, bool default_left, Operator cmp, int left_child_key, int right_child_key)=0
	Declare the current node as a numerical test node, where the test is of form [feature value] [cmp] [threshold]. Data points for which the test evaluates to True will be mapped to the left child node; all other data points (for which the test evaluates to False) will be mapped to the right child node. More...

virtual void	CategoricalTest (std::int32_t split_index, bool default_left, std::vector< std::uint32_t > const &category_list, bool category_list_right_child, int left_child_key, int right_child_key)=0
	Declare the current node as a categorical test node, where the test is of form [feature value] \in [category list]. More...

virtual void	LeafScalar (double leaf_value)=0
	Declare the current node as a leaf node with a scalar output. More...

virtual void	LeafVector (std::vector< float > const &leaf_vector)=0
	Declare the current node as a leaf node with a vector output. More...

virtual void	LeafVector (std::vector< double > const &leaf_vector)=0
	Declare the current node as a leaf node with a vector output. More...

virtual void	Gain (double gain)=0
	Specify the gain (loss reduction) that's resulted from the current split. More...

virtual void	DataCount (std::uint64_t data_count)=0
	Specify the number of data points (samples) that are mapped to the current node. More...

virtual void	SumHess (double sum_hess)=0
	Specify the weighted sample count or the sum of Hessians for the data points that are mapped to the current node. More...

virtual void	InitializeMetadata (Metadata const &metadata, TreeAnnotation const &tree_annotation, PostProcessorFunc const &postprocessor, std::vector< double > const &base_scores, std::optional< std::string > const &attributes)=0
	Specify a metadata for this model, if no metadata was previously specified. More...

virtual std::unique_ptr< Model >	CommitModel ()=0
	Conclude model building and obtain the final model object. More...

virtual	~ModelBuilder ()=default

Detailed Description

Model builder interface.

Note: A model builder object must be accessed by a single thread. For parallel tree construction, build multiple model objects and then concatenate them.

Constructor & Destructor Documentation

◆ ~ModelBuilder()

virtual treelite::model_builder::ModelBuilder::~ModelBuilder ( )

virtualdefault

Member Function Documentation

◆ CategoricalTest()

virtual void treelite::model_builder::ModelBuilder::CategoricalTest	(	std::int32_t	split_index,
		bool	default_left,
		std::vector< std::uint32_t > const &	category_list,
		bool	category_list_right_child,
		int	left_child_key,
		int	right_child_key
	)

pure virtual

Declare the current node as a categorical test node, where the test is of form [feature value] \in [category list].

Parameters

split_index	Feature ID
default_left	Whether the missing value should be mapped to the left child
category_list	List of categories to be tested for match
category_list_right_child	Whether the data points for which the test evaluates to True should be mapped to the right child or the left child.
left_child_key	Integer key that unique identifies the left child node.
right_child_key	Integer key that unique identifies the right child node.

◆ CommitModel()

virtual std::unique_ptr<Model> treelite::model_builder::ModelBuilder::CommitModel ( )

pure virtual

Conclude model building and obtain the final model object.

Returns: Final model object

◆ DataCount()

virtual void treelite::model_builder::ModelBuilder::DataCount ( std::uint64_t data_count )

pure virtual

Specify the number of data points (samples) that are mapped to the current node.

Parameters

data_count Number of data points

◆ EndNode()

virtual void treelite::model_builder::ModelBuilder::EndNode ( )

pure virtual

End the current node.

◆ EndTree()

virtual void treelite::model_builder::ModelBuilder::EndTree ( )

pure virtual

End the current tree.

◆ Gain()

virtual void treelite::model_builder::ModelBuilder::Gain ( double gain )

pure virtual

Specify the gain (loss reduction) that's resulted from the current split.

Parameters

gain	Gain (loss reduction)

◆ InitializeMetadata()

virtual void treelite::model_builder::ModelBuilder::InitializeMetadata	(	Metadata const &	metadata,
		TreeAnnotation const &	tree_annotation,
		PostProcessorFunc const &	postprocessor,
		std::vector< double > const &	base_scores,
		std::optional< std::string > const &	attributes
	)

pure virtual

Specify a metadata for this model, if no metadata was previously specified.

Parameters

metadata	Model metadata
tree_annotation	Annotation for individual trees
postprocessor	Postprocessor for prediction outputs
base_scores	Baseline scores for targets and classes, before adding tree outputs. Also known as the intercept.
attributes	Arbitrary JSON object, to be stored in the "attributes" field in the model object.

◆ LeafScalar()

virtual void treelite::model_builder::ModelBuilder::LeafScalar ( double leaf_value )

pure virtual

Declare the current node as a leaf node with a scalar output.

Parameters

leaf_value Value of leaf output

◆ LeafVector() [1/2]

virtual void treelite::model_builder::ModelBuilder::LeafVector ( std::vector< double > const & leaf_vector )

pure virtual

Declare the current node as a leaf node with a vector output.

Parameters

leaf_vector Value of leaf output

◆ LeafVector() [2/2]

virtual void treelite::model_builder::ModelBuilder::LeafVector ( std::vector< float > const & leaf_vector )

pure virtual

Declare the current node as a leaf node with a vector output.

Parameters

leaf_vector Value of leaf output

◆ NumericalTest()

virtual void treelite::model_builder::ModelBuilder::NumericalTest	(	std::int32_t	split_index,
		double	threshold,
		bool	default_left,
		Operator	cmp,
		int	left_child_key,
		int	right_child_key
	)

pure virtual

Declare the current node as a numerical test node, where the test is of form [feature value] [cmp] [threshold]. Data points for which the test evaluates to True will be mapped to the left child node; all other data points (for which the test evaluates to False) will be mapped to the right child node.

Parameters

split_index	Feature ID
threshold	Threshold
default_left	Whether the missing value should be mapped to the left child
cmp	Comparison operator
left_child_key	Integer key that unique identifies the left child node.
right_child_key	Integer key that unique identifies the right child node.

◆ StartNode()

virtual void treelite::model_builder::ModelBuilder::StartNode ( int node_key )

pure virtual

Start a new node.

Parameters

node_key Integer key that unique identifies the node.

◆ StartTree()

virtual void treelite::model_builder::ModelBuilder::StartTree ( )

pure virtual

Start a new tree.

◆ SumHess()

virtual void treelite::model_builder::ModelBuilder::SumHess ( double sum_hess )

pure virtual

Specify the weighted sample count or the sum of Hessians for the data points that are mapped to the current node.

Parameters

sum_hess Weighted sample count or the sum of Hessians

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

/home/docs/checkouts/readthedocs.org/user_builds/treelite/checkouts/latest/include/treelite/model_builder.h

Public Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ ~ModelBuilder()

Member Function Documentation

◆ CategoricalTest()

◆ CommitModel()

◆ DataCount()

◆ EndNode()

◆ EndTree()

◆ Gain()

◆ InitializeMetadata()

◆ LeafScalar()

◆ LeafVector() [1/2]

◆ LeafVector() [2/2]

◆ NumericalTest()

◆ StartNode()

◆ StartTree()

◆ SumHess()