jsat.clustering.dissimilarity

Class AbstractClusterDissimilarity

java.lang.Object

jsat.clustering.dissimilarity.AbstractClusterDissimilarity

All Implemented Interfaces:

ClusterDissimilarity

Direct Known Subclasses:

DistanceMetricDissimilarity

public abstract class AbstractClusterDissimilarity
extends Object
implements ClusterDissimilarity

This base class does not currently provide any inheritable functionality, but stores static methods.

Author:

Edward Raff

Constructor Summary

Constructors

Constructor and Description
AbstractClusterDissimilarity()

Method Summary

Modifier and Type	Method and Description
abstract ClusterDissimilarity	clone()
static double[][]	createDistanceMatrix(DataSet dataSet, ClusterDissimilarity cd) Creates an upper triangular matrix containing the distance between all points in the data set.
static double	getDistance(double[][] distanceMatrix, int i, int j) A convenience method.
static void	setDistance(double[][] distanceMatrix, int i, int j, double dist) A convenience method.

Methods inherited from class java.lang.Object

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

Methods inherited from interface jsat.clustering.dissimilarity.ClusterDissimilarity

dissimilarity, dissimilarity, distance

Constructor Detail

AbstractClusterDissimilarity

public AbstractClusterDissimilarity()

Method Detail

getDistance

public static double getDistance(double[][] distanceMatrix,
                                 int i,
                                 int j)

A convenience method. If the distanceMatrix was created with createDistanceMatrix(jsat.DataSet, jsat.clustering.dissimilarity.ClusterDissimilarity), then this method will return the appropriate value for the desired index.

Parameters:

distanceMatrix - the distance matrix to query from

i - the first index

j - the second index

Returns:

the correct value from the distance matrix from the index given as if the distance matrix was of full form

setDistance

public static void setDistance(double[][] distanceMatrix,
                               int i,
                               int j,
                               double dist)

A convenience method. If the distanceMatrix was created with createDistanceMatrix(jsat.DataSet, jsat.clustering.dissimilarity.ClusterDissimilarity), then this method will set the appropriate value for the desired index.

Parameters:

distanceMatrix - the distance matrix to query from

i - the first index

j - the second index

dist - the new distance value to store in the matrix

createDistanceMatrix

public static double[][] createDistanceMatrix(DataSet dataSet,
                                              ClusterDissimilarity cd)

Creates an upper triangular matrix containing the distance between all points in the data set. The main diagonal will contain all zeros, since the distance between a point and itself is always zero. This main diagonal is not stored, and is implicit
To save space, the matrix is staggered, and is of a size such that all elements to the left of the main diagonal are not present.
To compute the index into the returned array for the index [i][j], the values should be switched such that i ≥ j, and accessed as [i][j-i-1]

Parameters:

dataSet - the data set to create distance matrix for

cd - the cluster dissimilarity measure to use

Returns:

a upper triangular distance matrix

clone

public abstract ClusterDissimilarity clone()

Specified by:

clone in interface ClusterDissimilarity

Overrides:

clone in class Object