org.apache.ignite.ml.tree.randomforest.data.statistics

Class LeafValuesComputer<T>

java.lang.Object

org.apache.ignite.ml.tree.randomforest.data.statistics.LeafValuesComputer<T>

Type Parameters:

T - Type of leaf statistic.

All Implemented Interfaces:

Serializable

Direct Known Subclasses:

ClassifierLeafValuesComputer, RegressionLeafValuesComputer

public abstract class LeafValuesComputer<T>
extends Object
implements Serializable

Class containing logic of leaf values computing after building of all trees in random forest.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
LeafValuesComputer()

Method Summary

Modifier and Type	Method and Description
protected abstract void	addElementToLeafStatistic(T leafStatAggr, BootstrappedVector vec, int sampleId) Save vector to leaf statistic.
protected abstract double	computeLeafValue(T stat) Compute value from leaf based on statistics on labels corresponds to leaf.
protected abstract T	createLeafStatsAggregator(int sampleId) Creates an instance of leaf statistics aggregator in according to concrete algorithm based on RandomForest.
protected abstract T	mergeLeafStats(T leftStats, T rightStats) Merge statistics for same leafs.
void	setValuesForLeaves(ArrayList<TreeRoot> roots, Dataset<EmptyContext,BootstrappedDatasetPartition> dataset) Takes a list of all built trees and in one map-reduceImpurityStatistics step collect statistics for evaluating leaf-values for each tree and sets values for leaves.

Methods inherited from class java.lang.Object

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

Constructor Detail

LeafValuesComputer

public LeafValuesComputer()

Method Detail

setValuesForLeaves

public void setValuesForLeaves(ArrayList<TreeRoot> roots,
                               Dataset<EmptyContext,BootstrappedDatasetPartition> dataset)

Takes a list of all built trees and in one map-reduceImpurityStatistics step collect statistics for evaluating leaf-values for each tree and sets values for leaves.

Parameters:

roots - Learned trees.

dataset - Dataset.

addElementToLeafStatistic

protected abstract void addElementToLeafStatistic(T leafStatAggr,
                                                  BootstrappedVector vec,
                                                  int sampleId)

Save vector to leaf statistic.

Parameters:

leafStatAggr - Leaf statistics aggregator.

vec - Vector.

sampleId - Sample id.

mergeLeafStats

protected abstract T mergeLeafStats(T leftStats,
                                    T rightStats)

Merge statistics for same leafs.

Parameters:

leftStats - First leaf stat aggregator.

rightStats - Second leaf stat aggregator.

createLeafStatsAggregator

protected abstract T createLeafStatsAggregator(int sampleId)

Creates an instance of leaf statistics aggregator in according to concrete algorithm based on RandomForest.

Parameters:

sampleId - Sample id.

computeLeafValue

protected abstract double computeLeafValue(T stat)

Compute value from leaf based on statistics on labels corresponds to leaf.

Parameters:

stat - Leaf statistics.