mulan.evaluation.measure
Class LabelBasedAUC

java.lang.Object
  mulan.evaluation.measure.MeasureBase
      mulan.evaluation.measure.ConfidenceMeasureBase
          mulan.evaluation.measure.LabelBasedAUC

All Implemented Interfaces:

Serializable, Measure

Direct Known Subclasses:

MacroAUC, MicroAUC

public abstract class LabelBasedAUC
extends ConfidenceMeasureBase

Implementation of the label-based macro precision measure.

Version:

2012.07.17

Author:

Grigorios Tsoumakas

See Also:

Serialized Form

Field Summary

protected FastVector	all_Predictions The predictions for all labels
protected FastVector[]	m_Predictions The predictions for each label
protected int	numOfLabels The number of labels

Constructor Summary

LabelBasedAUC(int numOfLabels)
Creates a new instance of this class

Method Summary

double	getIdealValue() Gets an 'ideal' value of a measure.
void	reset() Resets the cumulated measure value, so the process of computation can be started from beginning (e.g. for a new series of outputs from learning task).
protected void	updateConfidence(double[] confidences, boolean[] truth) Updates the measure for a new example

Methods inherited from class mulan.evaluation.measure.ConfidenceMeasureBase

updateInternal

Methods inherited from class mulan.evaluation.measure.MeasureBase

makeCopy, toString, update

Methods inherited from class java.lang.Object

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

Methods inherited from interface mulan.evaluation.measure.Measure

getName, getValue

Field Detail

numOfLabels

protected int numOfLabels

The number of labels

m_Predictions

protected FastVector[] m_Predictions

The predictions for each label

all_Predictions

protected FastVector all_Predictions

The predictions for all labels

Constructor Detail

LabelBasedAUC

public LabelBasedAUC(int numOfLabels)

Creates a new instance of this class

Parameters:

numOfLabels - the number of labels

Method Detail

reset

public void reset()

Description copied from interface: Measure

Resets the cumulated measure value, so the process of computation can be started from beginning (e.g. for a new series of outputs from learning task).

getIdealValue

public double getIdealValue()

Description copied from interface: Measure

Gets an 'ideal' value of a measure. The 'ideal' means, that the value represents the best achievable performance of a learner for an prediction of a multi-label task and associated true labels.

Returns:

the ideal value

updateConfidence

protected void updateConfidence(double[] confidences,
                                boolean[] truth)

Description copied from class: ConfidenceMeasureBase

Updates the measure for a new example

Specified by:

updateConfidence in class ConfidenceMeasureBase

Parameters:

confidences - the confidences output by the learner for the example

truth - the ground truth of the example