moa.evaluation

Class CMM_GTAnalysis

java.lang.Object

moa.evaluation.CMM_GTAnalysis

public class CMM_GTAnalysis
extends Object

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
protected class	CMM_GTAnalysis.CMMPoint Wrapper class for data points to store CMM relevant attributes
protected class	CMM_GTAnalysis.GTCluster Main class to model the new clusters that will be the output of the cluster analysis

Constructor Summary

Constructors

Constructor and Description
CMM_GTAnalysis(Clustering trueClustering, ArrayList<DataPoint> points, boolean enableClassMerge)

Method Summary

Methods

Modifier and Type	Method and Description
double	getClassSeparability() Calculates how well the original clusters are separable.
protected double	getConnectionValue(CMM_GTAnalysis.CMMPoint cmmp, int clusterID) Calculate the connection of a point to a cluster
protected CMM_GTAnalysis.GTCluster	getGT0Cluster(int index) Return cluster
double	getModelQuality() Calculates the relative number of errors being caused by the underlying cluster model
double	getNoiseSeparability() Calculates how well noise is separable from the given clusters Small values indicate bad separability, values close to 1 indicate good separability
protected int	getNumberOfGT0Classes() Number of classes/clusters of the new clustering
String	getParameterString() String with main CMM parameters
protected CMM_GTAnalysis.CMMPoint	getPoint(int index) Get CMM internal point

Methods inherited from class java.lang.Object

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

Constructor Detail

CMM_GTAnalysis

public CMM_GTAnalysis(Clustering trueClustering,
              ArrayList<DataPoint> points,
              boolean enableClassMerge)

Parameters:

trueClustering - the ground truth clustering

points - data points

enableClassMerge - allow class merging (should be set to true on default)

Method Detail

getConnectionValue

protected double getConnectionValue(CMM_GTAnalysis.CMMPoint cmmp,
                        int clusterID)

Calculate the connection of a point to a cluster

Parameters:

cmmp - the point to calculate the connection for

clusterID - the corresponding cluster

Returns:

the connection value

getClassSeparability

public double getClassSeparability()

Calculates how well the original clusters are separable. Small values indicate bad separability, values close to 1 indicate good separability

Returns:

index of seperability

getNoiseSeparability

public double getNoiseSeparability()

Calculates how well noise is separable from the given clusters Small values indicate bad separability, values close to 1 indicate good separability

Returns:

index of noise separability

getModelQuality

public double getModelQuality()

Calculates the relative number of errors being caused by the underlying cluster model

Returns:

quality of the model

getPoint

protected CMM_GTAnalysis.CMMPoint getPoint(int index)

Get CMM internal point

Parameters:

index - of the point

Returns:

cmm point

getGT0Cluster

protected CMM_GTAnalysis.GTCluster getGT0Cluster(int index)

Return cluster

Parameters:

index - of the cluster to return

Returns:

cluster

getNumberOfGT0Classes

protected int getNumberOfGT0Classes()

Number of classes/clusters of the new clustering

Returns:

number of new clusters

getParameterString

public String getParameterString()

String with main CMM parameters

Returns:

main CMM parameter