edu.cmu.minorthird.classify.transform
Class ChiSquareInstanceTransform

java.lang.Object
  edu.cmu.minorthird.classify.transform.AbstractInstanceTransform
      edu.cmu.minorthird.classify.transform.ChiSquareInstanceTransform

All Implemented Interfaces:

InstanceTransform

public class ChiSquareInstanceTransform
extends AbstractInstanceTransform

Author:

Vitor R. Carvalho Date: March 2005 A simple feature filter based on Chi-squared statistic.

Constructor Summary

ChiSquareInstanceTransform()
Default Constructor - numFeatures=100

ChiSquareInstanceTransform(int num)
Constructor - parameter is number of features

Method Summary

void	addFeature(double infoGain, Feature f) Adds a score of feature f to the InstanceTransform
Feature[]	getTopFeatures(int top) get top ranked features
void	setNumberOfFeatures(int number) Number of features with the highest scores to keep in the dataset
java.lang.String	toString(int top) displays the top ranked features and their scores
Instance	transform(Instance ins) Transform an instance according to Info-Gain criterion

Methods inherited from class edu.cmu.minorthird.classify.transform.AbstractInstanceTransform

transform, transform, transform, transform

Methods inherited from class java.lang.Object

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

Constructor Detail

ChiSquareInstanceTransform

public ChiSquareInstanceTransform()

Default Constructor - numFeatures=100

ChiSquareInstanceTransform

public ChiSquareInstanceTransform(int num)

Constructor - parameter is number of features

Method Detail

transform

public Instance transform(Instance ins)

Transform an instance according to Info-Gain criterion

Specified by:

transform in interface InstanceTransform

Specified by:

transform in class AbstractInstanceTransform

toString

public java.lang.String toString(int top)

displays the top ranked features and their scores

getTopFeatures

public Feature[] getTopFeatures(int top)

get top ranked features

setNumberOfFeatures

public void setNumberOfFeatures(int number)

Number of features with the highest scores to keep in the dataset

addFeature

public void addFeature(double infoGain,
                       Feature f)

Adds a score of feature f to the InstanceTransform