org.apache.spark.mllib.tree.impl

Class DTStatsAggregator

Object

org.apache.spark.mllib.tree.impl.DTStatsAggregator

All Implemented Interfaces:

java.io.Serializable

public class DTStatsAggregator
extends Object
implements scala.Serializable

DecisionTree statistics aggregator for a node. This holds a flat array of statistics for a set of (features, bins) and helps with indexing. This class is abstract to support learning with and without feature subsampling.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
DTStatsAggregator(DecisionTreeMetadata metadata, scala.Option<int[]> featureSubset)

Method Summary

Methods

Modifier and Type	Method and Description
void	featureUpdate(int featureOffset, int binIndex, double label, double instanceWeight) Faster version of update.
int	getFeatureOffset(int featureIndex) Pre-compute feature offset for use with featureUpdate.
ImpurityCalculator	getImpurityCalculator(int featureOffset, int binIndex) Get an ImpurityCalculator for a given (node, feature, bin).
scala.Tuple2<Object,Object>	getLeftRightFeatureOffsets(int featureIndex) Pre-compute feature offset for use with featureUpdate.
ImpurityAggregator	impurityAggregator() ImpurityAggregator instance specifying the impurity type.
DTStatsAggregator	merge(DTStatsAggregator other) Merge this aggregator with another, and returns this aggregator.
void	mergeForFeature(int featureOffset, int binIndex, int otherBinIndex) For a given feature, merge the stats for two bins.
DecisionTreeMetadata	metadata()
void	update(int featureIndex, int binIndex, double label, double instanceWeight) Update the stats for a given (feature, bin) for ordered features, using the given label.

Methods inherited from class Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

DTStatsAggregator

public DTStatsAggregator(DecisionTreeMetadata metadata,
                 scala.Option<int[]> featureSubset)

Method Detail

metadata

public DecisionTreeMetadata metadata()

impurityAggregator

public ImpurityAggregator impurityAggregator()

ImpurityAggregator instance specifying the impurity type.

getImpurityCalculator

public ImpurityCalculator getImpurityCalculator(int featureOffset,
                                       int binIndex)

Get an ImpurityCalculator for a given (node, feature, bin).

Parameters:

featureOffset - For ordered features, this is a pre-computed (node, feature) offset from getFeatureOffset. For unordered features, this is a pre-computed (node, feature, left/right child) offset from getLeftRightFeatureOffsets.

update

public void update(int featureIndex,
          int binIndex,
          double label,
          double instanceWeight)

Update the stats for a given (feature, bin) for ordered features, using the given label.

featureUpdate

public void featureUpdate(int featureOffset,
                 int binIndex,
                 double label,
                 double instanceWeight)

Faster version of update. Update the stats for a given (feature, bin), using the given label.

Parameters:

featureOffset - For ordered features, this is a pre-computed feature offset from getFeatureOffset. For unordered features, this is a pre-computed (feature, left/right child) offset from getLeftRightFeatureOffsets.

getFeatureOffset

public int getFeatureOffset(int featureIndex)

Pre-compute feature offset for use with featureUpdate. For ordered features only.

getLeftRightFeatureOffsets

public scala.Tuple2<Object,Object> getLeftRightFeatureOffsets(int featureIndex)

Pre-compute feature offset for use with featureUpdate. For unordered features only.

mergeForFeature

public void mergeForFeature(int featureOffset,
                   int binIndex,
                   int otherBinIndex)

For a given feature, merge the stats for two bins.

Parameters:

featureOffset - For ordered features, this is a pre-computed feature offset from getFeatureOffset. For unordered features, this is a pre-computed (feature, left/right child) offset from getLeftRightFeatureOffsets.

binIndex - The other bin is merged into this bin.

otherBinIndex - This bin is not modified.

merge

public DTStatsAggregator merge(DTStatsAggregator other)

Merge this aggregator with another, and returns this aggregator. This method modifies this aggregator in-place.