The BestBinaryNumericSplit is a splitting function for decision trees that will exhaustively search a numeric dimension for the best binary split. More...

#include <best_binary_numeric_split.hpp>

Classes
class	AuxiliarySplitInfo

Static Public Member Functions
template<bool UseWeights, typename VecType , typename WeightVecType >
static double	SplitIfBetter (const double bestGain, const VecType &data, const arma::Row< size_t > &labels, const size_t numClasses, const WeightVecType &weights, const size_t minimumLeafSize, const double minimumGainSplit, arma::vec &splitInfo, AuxiliarySplitInfo &aux)
	Check if we can split a node. More...

template<bool UseWeights, typename VecType , typename ResponsesType , typename WeightVecType >
static std::enable_if< !HasOptimizedBinarySplitForms< FitnessFunction, UseWeights >::value, double >::type	SplitIfBetter (const double bestGain, const VecType &data, const ResponsesType &responses, const WeightVecType &weights, const size_t minimumLeafSize, const double minimumGainSplit, double &splitInfo, AuxiliarySplitInfo &aux)
	Check if we can split a node. More...

template<bool UseWeights, typename VecType , typename ResponsesType , typename WeightVecType >
static std::enable_if< HasOptimizedBinarySplitForms< FitnessFunction, UseWeights >::value, double >::type	SplitIfBetter (const double bestGain, const VecType &data, const ResponsesType &responses, const WeightVecType &weights, const size_t minimumLeafSize, const double minimumGainSplit, double &splitInfo, AuxiliarySplitInfo &)
	Check if we can split a node. More...

static size_t	NumChildren (const double &, const AuxiliarySplitInfo &)
	Returns 2, since the binary split always has two children.

template<typename ElemType >
static size_t	CalculateDirection (const ElemType &point, const double &splitInfo, const AuxiliarySplitInfo &)
	Given a point, calculate which child it should go to (left or right). More...

Detailed Description

template<typename FitnessFunction>
class mlpack::tree::BestBinaryNumericSplit< FitnessFunction >

The BestBinaryNumericSplit is a splitting function for decision trees that will exhaustively search a numeric dimension for the best binary split.

Template Parameters

FitnessFunction Fitness function to use to calculate gain.

Member Function Documentation

◆ CalculateDirection()

template<typename FitnessFunction >

template<typename ElemType >

size_t mlpack::tree::BestBinaryNumericSplit< FitnessFunction >::CalculateDirection	(	const ElemType &	point,
		const double &	splitInfo,
		const AuxiliarySplitInfo &
	)

static

Given a point, calculate which child it should go to (left or right).

Parameters

point	Point to calculate direction of.
splitInfo	Auxiliary information for the split.
*	(aux) Auxiliary information for the split (Unused).

◆ SplitIfBetter() [1/3]

template<typename FitnessFunction >

template<bool UseWeights, typename VecType , typename WeightVecType >

double mlpack::tree::BestBinaryNumericSplit< FitnessFunction >::SplitIfBetter	(	const double	bestGain,
		const VecType &	data,
		const arma::Row< size_t > &	labels,
		const size_t	numClasses,
		const WeightVecType &	weights,
		const size_t	minimumLeafSize,
		const double	minimumGainSplit,
		arma::vec &	splitInfo,
		AuxiliarySplitInfo &	aux
	)

static

Check if we can split a node.

If we can split a node in a way that improves on 'bestGain', then we return the improved gain. Otherwise we return the value 'bestGain'. If a split is made, then splitInfo and aux may be modified.

This overload is used only for classification tasks.

Parameters

bestGain	Best gain seen so far (we'll only split if we find gain better than this).
data	The dimension of data points to check for a split in.
labels	Labels for each point.
numClasses	Number of classes in the dataset.
weights	Weights associated with labels.
minimumLeafSize	Minimum number of points in a leaf node for splitting.
minimumGainSplit	Minimum gain split.
splitInfo	Stores split information on a successful split.
aux	Auxiliary split information, which may be modified on a successful split.

◆ SplitIfBetter() [2/3]

template<typename FitnessFunction >

template<bool UseWeights, typename VecType , typename ResponsesType , typename WeightVecType >

std::enable_if< HasOptimizedBinarySplitForms< FitnessFunction, UseWeights >::value, double >::type mlpack::tree::BestBinaryNumericSplit< FitnessFunction >::SplitIfBetter	(	const double	bestGain,
		const VecType &	data,
		const ResponsesType &	responses,
		const WeightVecType &	weights,
		const size_t	minimumLeafSize,
		const double	minimumGainSplit,
		double &	splitInfo,
		AuxiliarySplitInfo &	aux
	)

static

Check if we can split a node.

If we can split a node in a way that improves on 'bestGain', then we return the improved gain. Otherwise we return the value 'bestGain'. If a split is made, then splitInfo and aux may be modified.

This overload is used only for regression tasks.

Parameters

bestGain	Best gain seen so far (we'll only split if we find gain better than this).
data	The dimension of data points to check for a split in.
responses	Responses for each point.
weights	Weights associated with responses.
minimumLeafSize	Minimum number of points in a leaf node for splitting.
minimumGainSplit	Minimum gain split.
splitInfo	Stores split information on a successful split.
aux	Auxiliary split information, which may be modified on a successful split.

◆ SplitIfBetter() [3/3]

template<typename FitnessFunction >

template<bool UseWeights, typename VecType , typename ResponsesType , typename WeightVecType >

static std::enable_if< HasOptimizedBinarySplitForms<FitnessFunction, UseWeights>::value, double>::type mlpack::tree::BestBinaryNumericSplit< FitnessFunction >::SplitIfBetter	(	const double	bestGain,
		const VecType &	data,
		const ResponsesType &	responses,
		const WeightVecType &	weights,
		const size_t	minimumLeafSize,
		const double	minimumGainSplit,
		double &	splitInfo,
		AuxiliarySplitInfo &
	)

static

Check if we can split a node.

If we can split a node in a way that improves on 'bestGain', then we return the improved gain. Otherwise we return the value 'bestGain'. If a split is made, then splitInfo and aux may be modified.

This overload is specialized for any fitness function that implements BinaryScanInitialize(), BinaryStep() and BinaryGains() functions.

Parameters

bestGain	Best gain seen so far (we'll only split if we find gain better than this).
data	The dimension of data points to check for a split in.
responses	Responses for each point.
weights	Weights associated with responses.
minimumLeafSize	Minimum number of points in a leaf node for splitting.
minimumGainSplit	Minimum gain split.
splitInfo	Stores split information on a successful split.
aux	Auxiliary split information, which may be modified on a successful split.

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

src/mlpack/methods/decision_tree/best_binary_numeric_split.hpp
src/mlpack/methods/decision_tree/best_binary_numeric_split_impl.hpp

Classes

Static Public Member Functions

Detailed Description

template<typename FitnessFunction> class mlpack::tree::BestBinaryNumericSplit< FitnessFunction >

Member Function Documentation

◆ CalculateDirection()

◆ SplitIfBetter() [1/3]

◆ SplitIfBetter() [2/3]

◆ SplitIfBetter() [3/3]

template<typename FitnessFunction>
class mlpack::tree::BestBinaryNumericSplit< FitnessFunction >