weka.classifiers.evaluation

Class AggregateableEvaluationWithPriors

java.lang.Object

weka.classifiers.evaluation.Evaluation

weka.classifiers.evaluation.AggregateableEvaluation

weka.classifiers.evaluation.AggregateableEvaluationWithPriors

All Implemented Interfaces:

java.io.Serializable, weka.core.Aggregateable<weka.classifiers.evaluation.Evaluation>, weka.core.RevisionHandler, weka.core.Summarizable

public class AggregateableEvaluationWithPriors
extends weka.classifiers.evaluation.AggregateableEvaluation

A version of AggregateableEvaluation that allows priors to be provided (rather than being computed from the training data).

Version:

$Revision: 11274 $

Author:

Mark Hall (mhall{[at]}pentaho{[dot]}com)

See Also:

Serialized Form

Field Summary

Fields inherited from class weka.classifiers.evaluation.Evaluation

BUILT_IN_EVAL_METRICS

Constructor Summary

Constructors

Constructor and Description
AggregateableEvaluationWithPriors(weka.classifiers.evaluation.Evaluation eval) Constructs a new AggregateableEvaluationWithPriors based on another Evaluation object
AggregateableEvaluationWithPriors(weka.core.Instances data) Constructs a new AggregateableEvaluation object
AggregateableEvaluationWithPriors(weka.core.Instances data, weka.classifiers.CostMatrix costMatrix) Constructs a new AggregateableEvaluationWithPriors object

Method Summary

Modifier and Type	Method and Description
void	deleteStoredPredictions() Delete any buffered predictions
void	prunePredictions(double retain, long seed) Randomly downsample the predictions
void	setPriors(double[] priors, double count) Set the priors to use.

Methods inherited from class weka.classifiers.evaluation.AggregateableEvaluation

aggregate, finalizeAggregation

Methods inherited from class weka.classifiers.evaluation.Evaluation

areaUnderPRC, areaUnderROC, avgCost, confusionMatrix, correct, correlationCoefficient, coverageOfTestCasesByPredictedRegions, crossValidateModel, crossValidateModel, crossValidateModel, equals, errorRate, evaluateModel, evaluateModel, evaluateModel, evaluateModelOnce, evaluateModelOnce, evaluateModelOnce, evaluateModelOnceAndRecordPrediction, evaluateModelOnceAndRecordPrediction, evaluationForSingleInstance, falseNegativeRate, falseNegativeRate, falsePositiveRate, falsePositiveRate, fMeasure, fMeasure, getAllEvaluationMetricNames, getClassPriors, getDiscardPredictions, getHeader, getMetricsToDisplay, getPluginMetric, getPluginMetrics, getRevision, incorrect, kappa, KBInformation, KBMeanInformation, KBRelativeInformation, main, matthewsCorrelationCoefficient, matthewsCorrelationCoefficient, meanAbsoluteError, meanPriorAbsoluteError, missingClass, numClassified, numFalseNegatives, numFalsePositives, numInstances, numNegatives, numPositives, numPredictedNegatives, numPredictedPositives, numTrueNegatives, numTruePositives, pctCorrect, pctIncorrect, pctUnclassified, precision, predictedClassCounts, predictions, priorEntropy, recall, relativeAbsoluteError, rootMeanPriorSquaredError, rootMeanSquaredError, rootRelativeSquaredError, setDiscardPredictions, setMetricsToDisplay, setPriors, SFEntropyGain, SFMeanEntropyGain, SFMeanPriorEntropy, SFMeanSchemeEntropy, SFPriorEntropy, SFSchemeEntropy, sizeOfPredictedRegions, toClassDetailsString, toClassDetailsString, toCumulativeMarginDistributionString, toggleEvalMetrics, toMatrixString, toMatrixString, toSummaryString, toSummaryString, toSummaryString, totalCost, trueClassCounts, trueNegativeRate, trueNegativeRate, truePositiveRate, truePositiveRate, unclassified, unweightedMacroFmeasure, unweightedMicroFmeasure, updatePriors, useNoPriors, weightedAreaUnderPRC, weightedAreaUnderROC, weightedFalseNegativeRate, weightedFalsePositiveRate, weightedFMeasure, weightedMatthewsCorrelation, weightedPrecision, weightedRecall, weightedTrueNegativeRate, weightedTruePositiveRate, wekaStaticWrapper, withClass

Methods inherited from class java.lang.Object

getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AggregateableEvaluationWithPriors

public AggregateableEvaluationWithPriors(weka.core.Instances data)
                                  throws java.lang.Exception

Constructs a new AggregateableEvaluation object

Parameters:

data - the Instances to use

Throws:

java.lang.Exception - if a problem occurs

AggregateableEvaluationWithPriors

public AggregateableEvaluationWithPriors(weka.core.Instances data,
                                         weka.classifiers.CostMatrix costMatrix)
                                  throws java.lang.Exception

Constructs a new AggregateableEvaluationWithPriors object

Parameters:

data - the Instances to use

costMatrix - the cost matrix to use

Throws:

java.lang.Exception - if a problem occurs

AggregateableEvaluationWithPriors

public AggregateableEvaluationWithPriors(weka.classifiers.evaluation.Evaluation eval)
                                  throws java.lang.Exception

Constructs a new AggregateableEvaluationWithPriors based on another Evaluation object

Parameters:

eval - an Evaluation object

Throws:

java.lang.Exception - if a problem occurs

Method Detail

setPriors

public void setPriors(double[] priors,
                      double count)
               throws java.lang.Exception

Set the priors to use. In the case of classification it is the count of each class value; in the case of regression it is the sum of the target.

Parameters:

priors - the priors to use

count - the number of observations the priors were computed from

Throws:

java.lang.Exception - if a problem occurs

prunePredictions

public void prunePredictions(double retain,
                             long seed)
                      throws java.lang.Exception

Randomly downsample the predictions

Parameters:

retain - the fraction of the predictions to retain

seed - the random seed to use

Throws:

java.lang.Exception - if a problem occurs

deleteStoredPredictions

public void deleteStoredPredictions()

Delete any buffered predictions