jsat.clustering.evaluation.intra

Class SumOfSqrdPairwiseDistances

java.lang.Object

jsat.clustering.evaluation.intra.SumOfSqrdPairwiseDistances

All Implemented Interfaces:

IntraClusterEvaluation

public class SumOfSqrdPairwiseDistances
extends java.lang.Object
implements IntraClusterEvaluation

Evaluates a cluster's validity by computing the normalized sum of pairwise distances for all points in the cluster.
Note, the normalization value for each cluster is 1/(2 * n), where n is the number of points in each cluster.

For general distance metrics, this requires O(n²) work. The EuclideanDistance is a special case, and takes only O(n) work.

Constructor Summary

Constructors

Constructor and Description
SumOfSqrdPairwiseDistances() Creates a new evaluator that uses the Euclidean distance
SumOfSqrdPairwiseDistances(DistanceMetric dm) Creates a new cluster evaluator using the given distance metric
SumOfSqrdPairwiseDistances(SumOfSqrdPairwiseDistances toCopy) Copy constructor

Method Summary

Modifier and Type	Method and Description
SumOfSqrdPairwiseDistances	clone()
double	evaluate(int[] designations, DataSet dataSet, int clusterID) Evaluates the cluster represented by the given list of data points.
double	evaluate(java.util.List<DataPoint> dataPoints) Evaluates the cluster represented by the given list of data points.
DistanceMetric	getDistanceMetric()
void	setDistanceMetric(DistanceMetric dm) Sets the distance metric to be used whenever this object is called to evaluate a cluster

Methods inherited from class java.lang.Object

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

Constructor Detail

SumOfSqrdPairwiseDistances

public SumOfSqrdPairwiseDistances()

Creates a new evaluator that uses the Euclidean distance

SumOfSqrdPairwiseDistances

public SumOfSqrdPairwiseDistances(DistanceMetric dm)

Creates a new cluster evaluator using the given distance metric

Parameters:

dm - the distance metric to use

SumOfSqrdPairwiseDistances

public SumOfSqrdPairwiseDistances(SumOfSqrdPairwiseDistances toCopy)

Copy constructor

Parameters:

toCopy - the object to copy

Method Detail

setDistanceMetric

public void setDistanceMetric(DistanceMetric dm)

Sets the distance metric to be used whenever this object is called to evaluate a cluster

Parameters:

dm - the distance metric to use

getDistanceMetric

public DistanceMetric getDistanceMetric()

Returns:

the distance metric being used for evaluation

evaluate

public double evaluate(int[] designations,
                       DataSet dataSet,
                       int clusterID)

Description copied from interface: IntraClusterEvaluation

Evaluates the cluster represented by the given list of data points.

Specified by:

evaluate in interface IntraClusterEvaluation

Parameters:

designations - the array of cluster designations for the data set

dataSet - the full data set of all clusters

clusterID - the cluster id in the designations array to return the evaluation of

Returns:

the value in the range [0, Inf) that indicates how well formed the cluster is.

evaluate

public double evaluate(java.util.List<DataPoint> dataPoints)

Description copied from interface: IntraClusterEvaluation

Evaluates the cluster represented by the given list of data points.

Specified by:

evaluate in interface IntraClusterEvaluation

Parameters:

dataPoints - the data points that make up this cluster

Returns:

the value in the range [0, Inf) that indicates how well formed the cluster is.

clone

public SumOfSqrdPairwiseDistances clone()

Specified by:

clone in interface IntraClusterEvaluation

Overrides:

clone in class java.lang.Object