edu.uci.jforests.sample

Class RankingSample

java.lang.Object

edu.uci.jforests.sample.Sample

edu.uci.jforests.sample.RankingSample

public class RankingSample
extends Sample

Nested Class Summary

Nested classes/interfaces inherited from class edu.uci.jforests.sample.Sample

Sample.BinFreq

Field Summary

Fields

Modifier and Type	Field and Description
int	numQueries
int[]	queryBoundaries
int[]	queryIndices

Fields inherited from class edu.uci.jforests.sample.Sample

dataset, indicesInDataset, indicesInParentSample, size, targets, weights

Constructor Summary

Constructors

Constructor and Description
RankingSample(RankingDataset dataset)
RankingSample(RankingDataset dataset, int[] queryIndices, int[] queryBoundaries, int[] instances, double[] weights, double[] targets, int[] indicesInParentSample, int docCount, int queryCount)

Method Summary

Modifier and Type	Method and Description
double[]	evaluateByQuery(double[] predictions, RankingEvaluationMetric evaluationMetric)
RankingSample	getAugmentedSampleWithDocSampling(int times, double rate, java.util.Random rnd)
RankingSample	getClone()
RankingSample	getFilteredSubSample(java.util.List<java.lang.Integer> qids)
RankingSample	getOutOfSample(java.util.List<java.lang.Integer> qids) Creates a sample from queries that are in this sample and their ids are not listed in the input list of qids.
Sample	getRandomDocDistBiasedSubSample(double rate, java.util.Random rnd) Returns a random subsample which is biased with respect to number of documents per query.
RankingSample	getRandomSubSample(double rate, java.util.Random rnd)
RankingSample	getZeroFilteredSample()
void	printDocsPerQuery()

Methods inherited from class edu.uci.jforests.sample.Sample

evaluate, evaluate, getOutOfSample, getRandomTargetBiasedSubSample, isEmpty

Methods inherited from class java.lang.Object

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

Field Detail

numQueries

public int numQueries

queryBoundaries

public int[] queryBoundaries

queryIndices

public int[] queryIndices

Constructor Detail

RankingSample

public RankingSample(RankingDataset dataset)

RankingSample

public RankingSample(RankingDataset dataset,
                     int[] queryIndices,
                     int[] queryBoundaries,
                     int[] instances,
                     double[] weights,
                     double[] targets,
                     int[] indicesInParentSample,
                     int docCount,
                     int queryCount)

Method Detail

getRandomSubSample

public RankingSample getRandomSubSample(double rate,
                                        java.util.Random rnd)

Overrides:

getRandomSubSample in class Sample

getFilteredSubSample

public RankingSample getFilteredSubSample(java.util.List<java.lang.Integer> qids)

getOutOfSample

public RankingSample getOutOfSample(java.util.List<java.lang.Integer> qids)

Creates a sample from queries that are in this sample and their ids are not listed in the input list of qids.

getAugmentedSampleWithDocSampling

public RankingSample getAugmentedSampleWithDocSampling(int times,
                                                       double rate,
                                                       java.util.Random rnd)

getClone

public RankingSample getClone()

Overrides:

getClone in class Sample

evaluateByQuery

public double[] evaluateByQuery(double[] predictions,
                                RankingEvaluationMetric evaluationMetric)
                         throws java.lang.Exception

Throws:

java.lang.Exception

getZeroFilteredSample

public RankingSample getZeroFilteredSample()

getRandomDocDistBiasedSubSample

public Sample getRandomDocDistBiasedSubSample(double rate,
                                              java.util.Random rnd)

Returns a random subsample which is biased with respect to number of documents per query. It tries to include queries that have more diverse number of documents

printDocsPerQuery

public void printDocsPerQuery()