mulan.dimensionalityReduction
Class BinaryRelevanceAttributeEvaluator

java.lang.Object
  weka.attributeSelection.ASEvaluation
      mulan.dimensionalityReduction.BinaryRelevanceAttributeEvaluator

All Implemented Interfaces:

Serializable, AttributeEvaluator, CapabilitiesHandler, RevisionHandler

public class BinaryRelevanceAttributeEvaluator
extends ASEvaluation
implements AttributeEvaluator

Author:

George Traianos, Grigorios Tsoumakas

See Also:

Serialized Form

Nested Class Summary

class

BinaryRelevanceAttributeEvaluator.Rank a wrapper class for score-based attribute ranking

Constructor Summary

BinaryRelevanceAttributeEvaluator(ASEvaluation ase, MultiLabelInstances mlData, String combapp, String norm, String mode)

Method Summary

double	average(double[][] scoreList, int index) average score combination approach
void	buildEvaluator(Instances data) Not supported
static double[]	dl(double[] array) divide by length (dl) normalization
static double[]	dm(double[] array) divide by maximum (dm) normalization
double	evaluateAttribute(int attribute) Evaluates an attribute
double	highest(double[][] scoreList, int index) highest score combination approach
double	lowest(double[][] scoreList, int index) lowest score combination approach
static double	norm(double[] vector) calculates the norm of a vector
static void	normalize(double[] array) normalizes an array (in the range of [0,1])
int[]	order(int[] ranking) orders the ranking scores according to their attributes' original indices
int[]	rankAsc(double[] scores) returns a ranking of attributes (where each attribute is represented by its index)
int[]	rankDesc(double[] scores) returns a ranking of attributes (where each attribute is represented by its index)

Methods inherited from class weka.attributeSelection.ASEvaluation

forName, getCapabilities, getRevision, makeCopies, postProcess, runEvaluator

Methods inherited from class java.lang.Object

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

Constructor Detail

BinaryRelevanceAttributeEvaluator

public BinaryRelevanceAttributeEvaluator(ASEvaluation ase,
                                         MultiLabelInstances mlData,
                                         String combapp,
                                         String norm,
                                         String mode)

Parameters:

ase -

mlData -

combapp - combination approach mode ("max", "avg", "min")

norm - normalization mode ("dl", "dm", "none")

mode - scoring mode ("eval", "rank")

Method Detail

rankAsc

public int[] rankAsc(double[] scores)

returns a ranking of attributes (where each attribute is represented by its index)

Parameters:

scores - the attributes' scorelist

Returns:

an ascending ranking of the attributes, based on their scores

rankDesc

public int[] rankDesc(double[] scores)

returns a ranking of attributes (where each attribute is represented by its index)

Parameters:

scores - the attributes' scorelist

Returns:

a descending ranking of the attributes, based on their scores

order

public int[] order(int[] ranking)

orders the ranking scores according to their attributes' original indices

Parameters:

ranking - a rank table

Returns:

the order of the ranking scores

highest

public double highest(double[][] scoreList,
                      int index)

highest score combination approach

Parameters:

scoreList - all attributes' score lists

index - the index of a specific attribute's score

Returns:

the highest score achieved in any of the the input score lists

lowest

public double lowest(double[][] scoreList,
                     int index)

lowest score combination approach

Parameters:

scoreList - all attributes' score lists

index - the index of a specific attribute's score

Returns:

the lowest score achieved in all of the input score lists

average

public double average(double[][] scoreList,
                      int index)

average score combination approach

Parameters:

scoreList - all attributes' score lists

index - the index of a specific attribute's score

Returns:

the average score achieved in all the score lists

norm

public static double norm(double[] vector)

calculates the norm of a vector

Parameters:

vector - a numeric array (as a vector)

Returns:

the norm of the given vector

normalize

public static void normalize(double[] array)

normalizes an array (in the range of [0,1])

Parameters:

array - a numeric array

dl

public static double[] dl(double[] array)

divide by length (dl) normalization

Parameters:

array - a numeric array

Returns:

a dl normalized copy of array

dm

public static double[] dm(double[] array)

divide by maximum (dm) normalization

Parameters:

array - a numeric array

Returns:

a dm normalized copy of array

evaluateAttribute

public double evaluateAttribute(int attribute)
                         throws Exception

Evaluates an attribute

Specified by:

evaluateAttribute in interface AttributeEvaluator

Parameters:

attribute - the attribute index

Returns:

the evaluation

Throws:

Exception

buildEvaluator

public void buildEvaluator(Instances data)
                    throws Exception

Not supported

Specified by:

buildEvaluator in class ASEvaluation

Parameters:

data -

Throws:

Exception