Java* API Reference for Intel® Data Analytics Acceleration Library 2018 Update 2

List of all members
Parameter Class Reference

Base class for parameters of the gradient boosted trees classification training algorithm. More...

Detailed Description

Member Function Documentation

◆ getFeaturesPerNode()

long getFeaturesPerNode ( )

Returns number of features tried as possible splits by the gradient boosted trees classification training algorithm If 0 then sqrt(p) is used, where p is the total number of features.

Returns
Number of features

◆ getLambda()

double getLambda ( )

Returns lambda parameter of the training procedure (L2 regularization on weights lambda) Range: [0, inf). Default is 1

Returns
lambda value

◆ getMaxBins()

long getMaxBins ( )

Returns maximal number of discrete bins to bucket continuous features. Used with 'inexact' split finding method only. Default is 256. Increasing the number results in higher computation costs

Returns
Maximal number of discrete bins to bucket continuous features

◆ getMaxIterations()

long getMaxIterations ( )

Returns maximal number of iterations parameter of the gradient boosted trees training algorithm

Returns
Number of iterations

◆ getMaxTreeDepth()

long getMaxTreeDepth ( )

Returns maximal tree depth. Default is 0 (unlimited)

Returns
Maximal tree depth

◆ getMinBinSize()

long getMinBinSize ( )

Returns minimal number of observations in a bin. Default is 5 Used with 'inexact' split finding method only.

Returns
Minimal number of observations in a bin

◆ getMinObservationsInLeafNode()

long getMinObservationsInLeafNode ( )

Returns minimal number of samples per node. Default is 1

Returns
Minimal number of samples

◆ getMinSplitLoss()

double getMinSplitLoss ( )

Returns loss regularization parameter of the training procedure. Min loss reduction required to make a further partition on a leaf node of the tree. Range: [0, inf). Default is 0

Returns
minSplitLoss value

◆ getObservationsPerTreeFraction()

double getObservationsPerTreeFraction ( )

Returns fraction of observations used for a training of one tree, 0 to 1. Default is 1 (sampling without replacement)

Returns
Fraction of observations

◆ getShrinkage()

double getShrinkage ( )

Returns shrinkage parameter (learning rate) of the training procedure Scales the contribution of each tree by a factor (0, 1]. Default is 0.3

Returns
Shrinkage value

◆ getSplitMethod()

SplitMethod getSplitMethod ( )

Returns split finding method of the gradient boosted trees training algorithm

Returns
Split finding method

◆ setEngine()

void setEngine ( com.intel.daal.algorithms.engines.BatchBase  engine)

Sets the engine to be used by the algorithm

Parameters
engineto be used by the algorithm

◆ setFeaturesPerNode()

void setFeaturesPerNode ( long  value)

Sets the number of features tried as possible splits by gradient boosted trees classification training algorithm. If 0 then sqrt(p) is used, where p is the total number of features.

Parameters
valueNumber of features

◆ setLambda()

void setLambda ( double  value)

Sets lambda parameter value

Parameters
valuelambda value

◆ setMaxBins()

void setMaxBins ( long  value)

Sets maximal number of discrete bins to bucket continuous features. Used with 'inexact' split finding method only. Default is 256. Increasing the number results in higher computation costs

Parameters
valueMaximal number of discrete bins to bucket continuous features

◆ setMaxIterations()

void setMaxIterations ( long  n)

Sets maximal number of iterations for the gradient boosted trees training algorithm

Parameters
nNumber of iterations

◆ setMaxTreeDepth()

void setMaxTreeDepth ( long  value)

Sets maximal tree depth. Default is 0 (unlimited)

Parameters
valueMaximal tree depth

◆ setMinBinSize()

void setMinBinSize ( long  value)

Sets minimal number of observations in a bin. Default is 5 Used with 'inexact' split finding method only.

Parameters
valueMinimal number of observations in a bin

◆ setMinObservationsInLeafNode()

void setMinObservationsInLeafNode ( long  value)

Sets minimal number of samples per node. Default is 1

Parameters
valueMinimal number of samples

◆ setMinSplitLoss()

void setMinSplitLoss ( double  value)

Sets minSplitLoss parameter value

Parameters
valueminSplitLoss value

◆ setObservationsPerTreeFraction()

void setObservationsPerTreeFraction ( double  value)

Sets fraction of observations used for a training of one tree, 0 to 1

Parameters
valueFraction of observations

◆ setShrinkage()

void setShrinkage ( double  value)

Sets shrinkage parameter value

Parameters
valueShrinkage value

◆ setSplitMethod()

void setSplitMethod ( SplitMethod  splitMethod)

Sets split finding method of the gradient boosted trees training algorithm

Parameters
splitMethodSplit finding method

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

For more complete information about compiler optimizations, see our Optimization Notice.