Package org.moeaframework.util.clustering

Class KMeansPlusPlusClustering

java.lang.Object

org.moeaframework.util.clustering.KMeansPlusPlusClustering

All Implemented Interfaces:

Clustering

public class KMeansPlusPlusClustering
extends Object
implements Clustering

A wrapper for the Apache Commons Math K-means++ algorithm. The "++" term specifies a procedure for initializing the cluster centers before performing the k-means algorithm, which avoids theoretical shortcomings of the k-means algorithm that result in suboptimal clustering.

References:

1. https://en.wikipedia.org/wiki/K-means%2B%2B
2. https://en.wikipedia.org/wiki/K-means_clustering

Constructor Summary

Constructors

Constructor	Description
KMeansPlusPlusClustering(DistanceMeasure.ApacheDistanceMeasure<ClusterableSolution> distanceMeasure)	Constructs a new K-means++ clustering method.

Method Summary

Modifier and Type	Method	Description
List<Cluster>	cluster(int size, Iterable<ClusterableSolution> solutions)	Construct clusters using the points defined by the clusterable solutions.

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.moeaframework.util.clustering.Clustering

cluster, clusterObjectives, clusterObjectives, clusterVariables, clusterVariables, truncate

Constructor Details

KMeansPlusPlusClustering

public KMeansPlusPlusClustering(DistanceMeasure.ApacheDistanceMeasure<ClusterableSolution> distanceMeasure)

Constructs a new K-means++ clustering method.

Parameters:

distanceMeasure - the distance measure used to construct the cluster

Method Details

cluster

public List<Cluster> cluster(int size,
 Iterable<ClusterableSolution> solutions)

Description copied from interface: Clustering

Construct clusters using the points defined by the clusterable solutions.

Specified by:

cluster in interface Clustering

Parameters:

size - the number of clusters to construct

solutions - the clusterable solutions

Returns:

the clusters