medusa.georgios.enhanced_mcl

Interface DistanceMeasure

All Superinterfaces:

java.io.Serializable

All Known Implementing Classes:

AbstractSimilarity

public interface DistanceMeasure
extends java.io.Serializable

A distance measure is an algorithm to calculate the distance, similarity or correlation between two instances. There are three types of distance measures: distance, similarity and correlation measures. Some distance measures are normalized, i.e. in the interval [0,1], but this is not required by the interface.

See Also:

net.sf.javaml.distance.AbstractDistance, net.sf.javaml.distance.AbstractSimilarity, net.sf.javaml.distance.AbstractCorrelation

Method Summary

Modifier and Type	Method and Description
boolean	compare(double x, double y) Returns whether the first distance, similarity or correlation is better than the second distance, similarity or correlation.
double	measure(Instance x, Instance y) Calculates the distance between two instances.

Method Detail

measure

double measure(Instance x,
               Instance y)

Calculates the distance between two instances.

Parameters:

i - the first instance

j - the second instance

Returns:

the distance between the two instances

compare

boolean compare(double x,
                double y)

Returns whether the first distance, similarity or correlation is better than the second distance, similarity or correlation. Both values should bee calculated using the same measure. For similarity measures the higher the similarity the better the measure, for distance measures it is the lower the better and for correlation measure the absolute value must be higher.

Parameters:

x - the first distance, similarity or correlation

y - the second distance, similarity or correlation

Returns:

true if the first distance is better than the second, false in other cases.