Package elki.clustering.kmeans.quality

Class BayesianInformationCriterionXMeans

java.lang.Object

elki.clustering.kmeans.quality.AbstractKMeansQualityMeasure<NumberVector>

elki.clustering.kmeans.quality.BayesianInformationCriterionXMeans

All Implemented Interfaces:

KMeansQualityMeasure<NumberVector>

@Title("Bayesian Information Criterion (X-means Version)")
@Reference(authors="D. Pelleg, A. Moore",
           title="X-means: Extending K-means with Efficient Estimation on the Number of Clusters",
           booktitle="Proc. 17th Int. Conf. on Machine Learning (ICML 2000)",
           url="http://www.pelleg.org/shared/hp/download/xmeans.ps",
           bibkey="DBLP:conf/icml/PellegM00")
public class BayesianInformationCriterionXMeans
extends AbstractKMeansQualityMeasure<NumberVector>

Bayesian Information Criterion (BIC), also known as Schwarz criterion (SBC, SBIC) for the use with evaluating k-means results.

This version tries to be close to the version used in X-means, although people have argued that there are errors in this formulation.

Reference:

D. Pelleg, A. Moore:
X-means: Extending K-means with Efficient Estimation on the Number of Clusters
Proc. 17th Int. Conf. on Machine Learning (ICML 2000)

Since:

0.7.0

Author:

Tibor Goldschwendt, Erich Schubert

Constructor Summary

Constructors

Constructor	Description
BayesianInformationCriterionXMeans()

Method Summary

Modifier and Type	Method	Description
boolean	isBetter(double currentCost, double bestCost)	Compare two scores.
static double	logLikelihoodXMeans(Relation<? extends NumberVector> relation, Clustering<? extends MeanModel> clustering, NumberVectorDistance<?> distance)	Computes log likelihood of an entire clustering.
<V extends NumberVector> double	quality(Clustering<? extends MeanModel> clustering, NumberVectorDistance<? super V> distance, Relation<V> relation)	Calculates and returns the quality measure.

Methods inherited from class elki.clustering.kmeans.quality.AbstractKMeansQualityMeasure

logLikelihood, numberOfFreeParameters, numPoints, varianceContributionOfCluster

Methods inherited from class java.lang.Object

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

Constructor Detail

BayesianInformationCriterionXMeans

public BayesianInformationCriterionXMeans()

Method Detail

quality

public <V extends NumberVector> double quality(Clustering<? extends MeanModel> clustering,
                                               NumberVectorDistance<? super V> distance,
                                               Relation<V> relation)

Description copied from interface: KMeansQualityMeasure

Calculates and returns the quality measure.

Type Parameters:

V - Actual vector type (could be a subtype of O!)

Parameters:

clustering - Clustering to analyze

distance - Distance function to use (usually Euclidean or squared Euclidean!)

relation - Relation for accessing objects

Returns:

quality measure

logLikelihoodXMeans

public static double logLikelihoodXMeans(Relation<? extends NumberVector> relation,
                                         Clustering<? extends MeanModel> clustering,
                                         NumberVectorDistance<?> distance)

Computes log likelihood of an entire clustering.

Version as used in the X-means publication.

Parameters:

relation - Data relation

clustering - Clustering

distance - Distance function

Returns:

Log Likelihood.

isBetter

public boolean isBetter(double currentCost,
                        double bestCost)

Description copied from interface: KMeansQualityMeasure

Compare two scores.

Parameters:

currentCost - New (candiate) cost/score

bestCost - Existing best cost/score (may be NaN)

Returns:

true when the new score is better, or the old score is NaN.