Package smile.validation.metric

Class NormalizedMutualInformation

java.lang.Object

smile.validation.metric.NormalizedMutualInformation

All Implemented Interfaces:

Serializable, ClusteringMetric

public class NormalizedMutualInformation
extends Object
implements ClusteringMetric

Normalized Mutual Information (NMI) for comparing clustering. Normalized mutual information is between 0 (no mutual information) and 1 (perfect correlation). Normalization has been shown to improve the sensitiveness with respect to the difference in cluster distribution in the two clusterings.

References

1. X. Vinh, J. Epps, J. Bailey. Information Theoretic Measures for Clusterings Comparison: Variants, Properties, Normalization and Correction for Chance. JMLR, 2010.

See Also:

• Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static enum	NormalizedMutualInformation.Method	The normalization method.

Field Summary

Fields

Modifier and Type	Field	Description
static final NormalizedMutualInformation	JOINT	Default instance with max normalization.
static final NormalizedMutualInformation	MAX	Default instance with max normalization.
static final NormalizedMutualInformation	MIN	Default instance with min normalization.
static final NormalizedMutualInformation	SQRT	Default instance with sqrt normalization.
static final NormalizedMutualInformation	SUM	Default instance with sum normalization.

Constructor Summary

Constructors

Constructor	Description
NormalizedMutualInformation(NormalizedMutualInformation.Method method)	Constructor.

Method Summary

Modifier and Type	Method	Description
static double	joint(int[] y1, int[] y2)	Calculates the normalized mutual information of I(y1, y2) / H(y1, y2).
static double	max(int[] y1, int[] y2)	Calculates the normalized mutual information of I(y1, y2) / max(H(y1), H(y2)).
static double	min(int[] y1, int[] y2)	Calculates the normalized mutual information of I(y1, y2) / min(H(y1), H(y2)).
double	score(int[] y1, int[] y2)	Returns a score to measure the quality of clustering.
static double	sqrt(int[] y1, int[] y2)	Calculates the normalized mutual information of I(y1, y2) / sqrt(H(y1) * H(y2)).
static double	sum(int[] y1, int[] y2)	Calculates the normalized mutual information of 2 * I(y1, y2) / (H(y1) + H(y2)).
String	toString()

Methods inherited from class java.lang.Object

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

Field Details

JOINT

public static final NormalizedMutualInformation JOINT

Default instance with max normalization.

MAX

public static final NormalizedMutualInformation MAX

Default instance with max normalization.

MIN

public static final NormalizedMutualInformation MIN

Default instance with min normalization.

SUM

public static final NormalizedMutualInformation SUM

Default instance with sum normalization.

SQRT

public static final NormalizedMutualInformation SQRT

Default instance with sqrt normalization.

Constructor Details

NormalizedMutualInformation

public NormalizedMutualInformation(NormalizedMutualInformation.Method method)

Constructor.

Parameters:

method - the normalization method.

Method Details

score

public double score(int[] y1,
 int[] y2)

Description copied from interface: ClusteringMetric

Returns a score to measure the quality of clustering.

Specified by:

score in interface ClusteringMetric

Parameters:

y1 - the ground truth (or simply a clustering labels).

y2 - the alternative cluster labels.

Returns:

the metric.

joint

public static double joint(int[] y1,
 int[] y2)

Calculates the normalized mutual information of I(y1, y2) / H(y1, y2).

Parameters:

y1 - the clustering labels.

y2 - the alternative cluster labels.

Returns:

the metric.

max

public static double max(int[] y1,
 int[] y2)

Calculates the normalized mutual information of I(y1, y2) / max(H(y1), H(y2)).

Parameters:

y1 - the clustering labels.

y2 - the alternative cluster labels.

Returns:

the metric.

sum

public static double sum(int[] y1,
 int[] y2)

Calculates the normalized mutual information of 2 * I(y1, y2) / (H(y1) + H(y2)).

Parameters:

y1 - the clustering labels.

y2 - the alternative cluster labels.

Returns:

the metric.

sqrt

public static double sqrt(int[] y1,
 int[] y2)

Calculates the normalized mutual information of I(y1, y2) / sqrt(H(y1) * H(y2)).

Parameters:

y1 - the clustering labels.

y2 - the alternative cluster labels.

Returns:

the metric.

min

public static double min(int[] y1,
 int[] y2)

Calculates the normalized mutual information of I(y1, y2) / min(H(y1), H(y2)).

Parameters:

y1 - the clustering labels.

y2 - the alternative cluster labels.

Returns:

the metric.

toString

public String toString()

Overrides:

toString in class Object