com.rapidminer.operator.clustering.clusterer
Class AbstractLinkageMethod

java.lang.Object
  com.rapidminer.operator.clustering.clusterer.AbstractLinkageMethod

Direct Known Subclasses:

AverageLinkageMethod, CompleteLinkageMethod, SingleLinkageMethod

public abstract class AbstractLinkageMethod
extends java.lang.Object

This class provides the basic functionality for all linkage methods of agglomerative clustering. It stores a distance matrix between all clusters and returns the next agglomeration as the minimum of all distances. To save time needed to copy the matrix if two clusters are joined, it is not resized, instead one row and column is not used anymore. The other row and column are updated by the agglomeration methods.

Author:

Sebastian Land

Constructor Summary

AbstractLinkageMethod(DistanceMatrix matrix, int[] clusterIds)

Method Summary

Agglomeration	getNextAgglomeration(int nextClusterId, java.util.Map<java.lang.Integer,HierarchicalClusterNode> clusterMap)
abstract void	updateDistances(DistanceMatrix matrix, int updatedRow, int unionedRow, java.util.Map<java.lang.Integer,HierarchicalClusterNode> clusterMap)

Methods inherited from class java.lang.Object

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

Constructor Detail

AbstractLinkageMethod

public AbstractLinkageMethod(DistanceMatrix matrix,
                             int[] clusterIds)

Method Detail

getNextAgglomeration

public Agglomeration getNextAgglomeration(int nextClusterId,
                                          java.util.Map<java.lang.Integer,HierarchicalClusterNode> clusterMap)

updateDistances

public abstract void updateDistances(DistanceMatrix matrix,
                                     int updatedRow,
                                     int unionedRow,
                                     java.util.Map<java.lang.Integer,HierarchicalClusterNode> clusterMap)