Package si.ijs.kt.clus.ext.semisupervised.confidence

Class OracleScore

java.lang.Object

si.ijs.kt.clus.ext.semisupervised.confidence.PredictionConfidence

si.ijs.kt.clus.ext.semisupervised.confidence.OracleScore

public class OracleScore
extends PredictionConfidence

Provides reliability scores on the basis of 'actual error', which is not attainable in practice, i.e., if true unlabeled data are used. The unlabeled instances provided need to have labels.

Author:

jurical

Field Summary

Fields inherited from class si.ijs.kt.clus.ext.semisupervised.confidence.PredictionConfidence

m_Aggregation, m_ConfidenceScores, m_counter, m_NbTargets, m_Normalization, m_perTargetScores, m_StatManager

Constructor Summary

Constructors

Constructor	Description
OracleScore(ClusStatManager statManager, SettingsSSL.SSLNormalization normalizationType, SettingsSSL.SSLAggregation aggregationType)

Method Summary

Modifier and Type	Method	Description
double[]	calculatePerTargetOOBScores(ClusForest model, DataTuple tuple)	Per-target reliability scores for out-of-bag (labeled) examples.
double[]	calculatePerTargetScores(ClusModel model, DataTuple tuple)	Returns per-target confidence scores
void	setHmcThreshold(double t)	Threshold for prediction in the context of hierarchical multi-label classification

Methods inherited from class si.ijs.kt.clus.ext.semisupervised.confidence.PredictionConfidence

calculateConfidenceScores, calculateConfidenceScores, calculateOOBConfidenceScores, getConfidence, getConfidenceScores, getCounter, getNbTargetAttributes, getNormalization, getPerTargetConfidenceScores

Methods inherited from class java.lang.Object

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

Constructor Details

OracleScore

public OracleScore(ClusStatManager statManager,
 SettingsSSL.SSLNormalization normalizationType,
 SettingsSSL.SSLAggregation aggregationType)

Method Details

calculatePerTargetScores

public double[] calculatePerTargetScores(ClusModel model,
 DataTuple tuple)
                                  throws ClusException,
InterruptedException

Returns per-target confidence scores

Specified by:

calculatePerTargetScores in class PredictionConfidence

Parameters:

model -

tuple - Tuple is assumed to have true labels

Throws:

ClusException

InterruptedException

calculatePerTargetOOBScores

public double[] calculatePerTargetOOBScores(ClusForest model,
 DataTuple tuple)

Description copied from class: PredictionConfidence

Per-target reliability scores for out-of-bag (labeled) examples. Applicable only for ClusForest.

Specified by:

calculatePerTargetOOBScores in class PredictionConfidence

setHmcThreshold

public void setHmcThreshold(double t)

Threshold for prediction in the context of hierarchical multi-label classification

Parameters:

t -