org.apache.ctakes.temporal.ae.feature.selection

Class BinaryAlphaFeatureSelection<OUTCOME_T>

java.lang.Object

org.apache.ctakes.temporal.ae.feature.selection.FeatureSelection<OUTCOME_T>

org.apache.ctakes.temporal.ae.feature.selection.BinaryAlphaFeatureSelection<OUTCOME_T>

public class BinaryAlphaFeatureSelection<OUTCOME_T>
extends FeatureSelection<OUTCOME_T>

Selects features via binary Krippendorff's Alpha-Reliability between the a extracted feature and the outcome values they are paired with in classification instances. Alpha-positive: when the positive feature value suggests positive label Alpha-negative: when the positive feature value suggests negative label Final Alpha = max(Alpha-positive, Alpha-negative) Reference: Krippendorff, Klaus. "Computing Krippendorff's alpha reliability." Departmental Papers (ASC) (2007): 43.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
private static class	BinaryAlphaFeatureSelection.AlphaScorer<OUTCOME_T> Helper class for aggregating and computing mutual Odds Ratio statistics

Field Summary

Fields

Modifier and Type	Field and Description
private BinaryAlphaFeatureSelection.AlphaScorer<OUTCOME_T>	alphaFunction
private java.util.LinkedHashSet<java.lang.String>	discardedFeatureNames
private double	featureSelectionThreshold
private int	numFeatures
private java.lang.String	positiveClass

Fields inherited from class org.apache.ctakes.temporal.ae.feature.selection.FeatureSelection

isTrained, selectedFeatureNames

Constructor Summary

Constructors

Constructor and Description
BinaryAlphaFeatureSelection(java.lang.String name)
BinaryAlphaFeatureSelection(java.lang.String name, double threshold)
BinaryAlphaFeatureSelection(java.lang.String name, double threshold, java.lang.String posiClas)

Method Summary

Methods

Modifier and Type	Method and Description
boolean	apply(Feature feature)
void	load(java.net.URI uri)
void	save(java.net.URI uri)
void	train(java.lang.Iterable<> instances)

Methods inherited from class org.apache.ctakes.temporal.ae.feature.selection.FeatureSelection

getFeatureName, transform, transform

Methods inherited from class java.lang.Object

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

Field Detail

featureSelectionThreshold

private double featureSelectionThreshold

numFeatures

private int numFeatures

positiveClass

private java.lang.String positiveClass

alphaFunction

private BinaryAlphaFeatureSelection.AlphaScorer<OUTCOME_T> alphaFunction

discardedFeatureNames

private java.util.LinkedHashSet<java.lang.String> discardedFeatureNames

Constructor Detail

BinaryAlphaFeatureSelection

public BinaryAlphaFeatureSelection(java.lang.String name)

BinaryAlphaFeatureSelection

public BinaryAlphaFeatureSelection(java.lang.String name,
                           double threshold)

BinaryAlphaFeatureSelection

public BinaryAlphaFeatureSelection(java.lang.String name,
                           double threshold,
                           java.lang.String posiClas)

Parameters:

name: - feature selection method name

threshold: - percentage threshold to control how many features to keep

posiClas: - specify which class is the positive class

Method Detail

apply

public boolean apply(Feature feature)

Overrides:

apply in class FeatureSelection<OUTCOME_T>

train

public void train(java.lang.Iterable<> instances)

save

public void save(java.net.URI uri)
          throws java.io.IOException

Throws:

java.io.IOException

load

public void load(java.net.URI uri)
          throws java.io.IOException

Throws:

java.io.IOException