edu.ucla.sspace.matrix

Class LogEntropyTransform.LogEntropyGlobalTransform

java.lang.Object

edu.ucla.sspace.matrix.LogEntropyTransform.LogEntropyGlobalTransform

All Implemented Interfaces:

GlobalTransform

Enclosing class:

LogEntropyTransform

public class LogEntropyTransform.LogEntropyGlobalTransform
extends Object
implements GlobalTransform

The real implementation of the Log Entropy transformation as a GlobalTransform

Constructor Summary

Constructors

Constructor and Description
LogEntropyTransform.LogEntropyGlobalTransform(File inputMatrixFile, MatrixIO.Format format) Creates an instance of LogEntropyGlobalTransform from a File of format Format.
LogEntropyTransform.LogEntropyGlobalTransform(Matrix matrix) Creates an instance of LogEntropyGlobalTransform from a Matrix.

Method Summary

Methods

Modifier and Type	Method and Description
double	transform(int row, int column, double value) Calculates the entropy (information gain) where value is the number of occurances of item row with feature column.

Methods inherited from class java.lang.Object

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

Constructor Detail

LogEntropyTransform.LogEntropyGlobalTransform

public LogEntropyTransform.LogEntropyGlobalTransform(Matrix matrix)

Creates an instance of LogEntropyGlobalTransform from a Matrix.

LogEntropyTransform.LogEntropyGlobalTransform

public LogEntropyTransform.LogEntropyGlobalTransform(File inputMatrixFile,
                                             MatrixIO.Format format)

Creates an instance of LogEntropyGlobalTransform from a File of format Format.

Method Detail

transform

public double transform(int row,
               int column,
               double value)

Calculates the entropy (information gain) where value is the number of occurances of item row with feature column. The item entropy is defined as:

1 + entropy(item) / log(numberOfFeatures)

with entropy defined as:

sum_features(p(item, feature) * log(p(item, feature)))

Specified by:

transform in interface GlobalTransform

Parameters:

row - The index specifying the observed item

column - The index specifying the observed feature

value - The number occurances of the item and the feature

Returns:

log(value) * item_entropy(row)