org.carrot2.filter.trc.carrot.filter.cluster.rough.clustering
Class AbstractKMeansAlgorithm

java.lang.Object
  org.carrot2.filter.trc.carrot.filter.cluster.rough.clustering.AbstractKMeansAlgorithm

Direct Known Subclasses:

SoftKMeansAlgorithm

public abstract class AbstractKMeansAlgorithm
extends Object

A boiler-plate implementation of generic K-Means algorithm. Soft (fuzzy) membership allowed.

Field Summary

protected static Timer

timer

Constructor Summary

AbstractKMeansAlgorithm()

Method Summary

protected abstract void	calculateClusterRepresentation() Calculate representation for cluster (usually in form of a centroid)
void	cluster() Execute clustering
protected abstract void	doClustering() Main step in clustering algorithm : assign objects to clusters
abstract double[][]	getClusterRepresentation() Return cluster's representation vector as a matrix Rows = vectors
abstract int[][]	getClusters() Return clusters as a form of binary matrix (n rows-objects * k colums-clusters) Cell[i,j] = 1 iff object i belongs to cluster j
abstract double[][]	getMembership() Return cluster membership matrix Colums = cluster membership vector
abstract int[]	getUnclassified() Return indices of unclassified objects
abstract double[]	getUnclassifiedCentroid() Return centroid of unclassified objects
abstract double[]	getUnclassifiedMembership() Return membership vector for unclassified objects; Length of vector = number of unclassified object
protected abstract void	prepareInitialClusters() Prepare initial clusters
protected abstract boolean	stopCondition() Stop condition for algorithm

Methods inherited from class java.lang.Object

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

Field Detail

timer

protected static final Timer timer

Constructor Detail

AbstractKMeansAlgorithm

public AbstractKMeansAlgorithm()

Method Detail

prepareInitialClusters

protected abstract void prepareInitialClusters()

Prepare initial clusters

doClustering

protected abstract void doClustering()

Main step in clustering algorithm : assign objects to clusters

calculateClusterRepresentation

protected abstract void calculateClusterRepresentation()

Calculate representation for cluster (usually in form of a centroid)

stopCondition

protected abstract boolean stopCondition()

Stop condition for algorithm

Returns:

true when algorithm should stop processing

cluster

public void cluster()

Execute clustering

getClusters

public abstract int[][] getClusters()

Return clusters as a form of binary matrix (n rows-objects * k colums-clusters) Cell[i,j] = 1 iff object i belongs to cluster j

getMembership

public abstract double[][] getMembership()

Return cluster membership matrix Colums = cluster membership vector

getClusterRepresentation

public abstract double[][] getClusterRepresentation()

Return cluster's representation vector as a matrix Rows = vectors

getUnclassified

public abstract int[] getUnclassified()

Return indices of unclassified objects

getUnclassifiedCentroid

public abstract double[] getUnclassifiedCentroid()

Return centroid of unclassified objects

getUnclassifiedMembership

public abstract double[] getUnclassifiedMembership()

Return membership vector for unclassified objects; Length of vector = number of unclassified object