moa.classifiers.meta

Class AccuracyWeightedEnsemble

java.lang.Object

moa.AbstractMOAObject

moa.options.AbstractOptionHandler

moa.classifiers.AbstractClassifier

moa.classifiers.meta.AccuracyWeightedEnsemble

All Implemented Interfaces:

Serializable, Classifier, AWTRenderable, MOAObject, OptionHandler

public class AccuracyWeightedEnsemble
extends AbstractClassifier

The Accuracy Weighted Ensemble classifier as proposed by Wang et al. in "Mining concept-drifting data streams using ensemble classifiers", KDD 2003.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected Classifier	candidateClassifier
protected int	chunkSize
IntOption	chunkSizeOption Chunk size.
protected long[]	classDistributions
protected weka.core.Instances	currentChunk
protected Classifier[]	ensemble
protected double[]	ensembleWeights
ClassOption	learnerOption Type of classifier to use as a component classifier.
protected int	maxMemberCount
protected int	maxStoredCount
FloatOption	memberCountOption Number of component classifiers.
protected int	numFolds
IntOption	numFoldsOption Number of folds in candidate classifier cross-validation.
protected int	processedInstances
FloatOption	storedCountOption Number of classifiers remembered and available for ensemble construction.
protected Classifier[]	storedLearners
protected double[][]	storedWeights The weights of stored classifiers.
protected static Comparator<double[]>	weightComparator Simple weight comparator.

Fields inherited from class moa.classifiers.AbstractClassifier

classifierRandom, modelContext, randomSeed, randomSeedOption, trainingWeightSeenByModel

Fields inherited from class moa.options.AbstractOptionHandler

classOptionNamesToPreparedObjects, options

Constructor Summary

Constructors

Constructor and Description
AccuracyWeightedEnsemble()

Method Summary

Methods

Modifier and Type	Method and Description
protected Classifier	addToStored(Classifier newClassifier, double newClassifiersWeight) Adds a classifier to the storage.
protected double	computeCandidateWeight(Classifier candidate, weka.core.Instances chunk, int numFolds) Computes the weight of a candidate classifier.
protected double	computeMseR() Computes the MSEr threshold.
protected double	computeWeight(Classifier learner, weka.core.Instances chunk) Computes the weight of a given classifie.
protected void	discardModel(int index) Removes the classifier at a given index from the model, thus decreasing the models size.
void	getModelDescription(StringBuilder out, int indent) Returns a string representation of the model.
protected Measurement[]	getModelMeasurementsImpl() Adds ensemble weights to the measurements.
String	getPurposeString() Gets the purpose of this object
Classifier[]	getSubClassifiers() Gets the classifiers of this ensemble.
double[]	getVotesForInstance(weka.core.Instance inst) Predicts a class for an example.
boolean	isRandomizable() Determines whether the classifier is randomizable.
void	prepareForUseImpl(TaskMonitor monitor, ObjectRepository repository) This method describes the implementation of how to prepare this object for use.
protected void	processChunk() Processes a chunk.
protected int	removePoorestModelBytes() Removes the poorest classifier from the model, thus decreasing the models size.
void	resetLearningImpl() Resets this classifier.
void	trainOnInstanceImpl(weka.core.Instance inst) Trains this classifier incrementally using the given instance. The reason for ...Impl methods: ease programmer burden by not requiring them to remember calls to super in overridden methods.

Methods inherited from class moa.classifiers.AbstractClassifier

contextIsCompatible, copy, correctlyClassifies, getAttributeNameString, getAWTRenderer, getClassLabelString, getClassNameString, getDescription, getModelContext, getModelMeasurements, getNominalValueString, modelAttIndexToInstanceAttIndex, modelAttIndexToInstanceAttIndex, resetLearning, setModelContext, setRandomSeed, trainingHasStarted, trainingWeightSeenByModel, trainOnInstance

Methods inherited from class moa.options.AbstractOptionHandler

discoverOptionsViaReflection, getCLICreationString, getOptions, getPreparedClassOption, getPreparedClassOption, prepareClassOptions, prepareForUse, prepareForUse

Methods inherited from class moa.AbstractMOAObject

copy, measureByteSize, measureByteSize, toString

Methods inherited from class java.lang.Object

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

Methods inherited from interface moa.options.OptionHandler

getCLICreationString, getOptions, prepareForUse, prepareForUse

Methods inherited from interface moa.MOAObject

measureByteSize

Field Detail

weightComparator

protected static Comparator<double[]> weightComparator

Simple weight comparator.

learnerOption

public ClassOption learnerOption

Type of classifier to use as a component classifier.

memberCountOption

public FloatOption memberCountOption

Number of component classifiers.

storedCountOption

public FloatOption storedCountOption

Number of classifiers remembered and available for ensemble construction.

chunkSizeOption

public IntOption chunkSizeOption

Chunk size.

numFoldsOption

public IntOption numFoldsOption

Number of folds in candidate classifier cross-validation.

classDistributions

protected long[] classDistributions

ensemble

protected Classifier[] ensemble

storedLearners

protected Classifier[] storedLearners

ensembleWeights

protected double[] ensembleWeights

storedWeights

protected double[][] storedWeights

The weights of stored classifiers. storedWeights[x][0] = weight storedWeights[x][1] = classifier

processedInstances

protected int processedInstances

chunkSize

protected int chunkSize

numFolds

protected int numFolds

maxMemberCount

protected int maxMemberCount

maxStoredCount

protected int maxStoredCount

candidateClassifier

protected Classifier candidateClassifier

currentChunk

protected weka.core.Instances currentChunk

Constructor Detail

AccuracyWeightedEnsemble

public AccuracyWeightedEnsemble()

Method Detail

getPurposeString

public String getPurposeString()

Description copied from interface: OptionHandler

Gets the purpose of this object

Specified by:

getPurposeString in interface OptionHandler

Overrides:

getPurposeString in class AbstractClassifier

Returns:

the string with the purpose of this object

prepareForUseImpl

public void prepareForUseImpl(TaskMonitor monitor,
                     ObjectRepository repository)

Description copied from class: AbstractOptionHandler

This method describes the implementation of how to prepare this object for use. All classes that extends this class have to implement prepareForUseImpl and not prepareForUse since prepareForUse calls prepareForUseImpl.

Overrides:

prepareForUseImpl in class AbstractClassifier

Parameters:

monitor - the TaskMonitor to use

repository - the ObjectRepository to use

resetLearningImpl

public void resetLearningImpl()

Description copied from class: AbstractClassifier

Resets this classifier. It must be similar to starting a new classifier from scratch.

The reason for ...Impl methods: ease programmer burden by not requiring them to remember calls to super in overridden methods. Note that this will produce compiler errors if not overridden.

Specified by:

resetLearningImpl in class AbstractClassifier

trainOnInstanceImpl

public void trainOnInstanceImpl(weka.core.Instance inst)

Description copied from class: AbstractClassifier

Trains this classifier incrementally using the given instance.

The reason for ...Impl methods: ease programmer burden by not requiring them to remember calls to super in overridden methods. Note that this will produce compiler errors if not overridden.

Specified by:

trainOnInstanceImpl in class AbstractClassifier

Parameters:

inst - the instance to be used for training

processChunk

protected void processChunk()

Processes a chunk.

Parameters:

useMseR - Determines whether to use the MSEr threshold.

computeCandidateWeight

protected double computeCandidateWeight(Classifier candidate,
                            weka.core.Instances chunk,
                            int numFolds)

Computes the weight of a candidate classifier.

Parameters:

candidate - Candidate classifier.

chunk - Data chunk of examples.

numFolds - Number of folds in candidate classifier cross-validation.

useMseR - Determines whether to use the MSEr threshold.

Returns:

Candidate classifier weight.

computeWeight

protected double computeWeight(Classifier learner,
                   weka.core.Instances chunk)

Computes the weight of a given classifie.

Parameters:

learner - Classifier to calculate weight for.

chunk - Data chunk of examples.

useMseR - Determines whether to use the MSEr threshold.

Returns:

The given classifier's weight.

computeMseR

protected double computeMseR()

Computes the MSEr threshold.

Returns:

The MSEr threshold.

getVotesForInstance

public double[] getVotesForInstance(weka.core.Instance inst)

Predicts a class for an example.

Parameters:

inst - the instance to be classified

Returns:

an array containing the estimated membership probabilities of the test instance in each class

getModelDescription

public void getModelDescription(StringBuilder out,
                       int indent)

Description copied from class: AbstractClassifier

Returns a string representation of the model.

Specified by:

getModelDescription in class AbstractClassifier

Parameters:

out - the stringbuilder to add the description

indent - the number of characters to indent

getModelMeasurementsImpl

protected Measurement[] getModelMeasurementsImpl()

Adds ensemble weights to the measurements.

Specified by:

getModelMeasurementsImpl in class AbstractClassifier

Returns:

an array of measurements to be used in evaluation tasks

isRandomizable

public boolean isRandomizable()

Determines whether the classifier is randomizable.

Returns:

true if the classifier needs a random seed.

getSubClassifiers

public Classifier[] getSubClassifiers()

Description copied from interface: Classifier

Gets the classifiers of this ensemble. Returns null if this classifier is a single classifier.

Specified by:

getSubClassifiers in interface Classifier

Overrides:

getSubClassifiers in class AbstractClassifier

Returns:

an array of the classifiers of the ensemble

addToStored

protected Classifier addToStored(Classifier newClassifier,
                     double newClassifiersWeight)

Adds a classifier to the storage.

Parameters:

newClassifier - The classifier to add.

newClassifiersWeight - The new classifiers weight.

removePoorestModelBytes

protected int removePoorestModelBytes()

Removes the poorest classifier from the model, thus decreasing the models size.

Returns:

the size of the removed classifier.

discardModel

protected void discardModel(int index)

Removes the classifier at a given index from the model, thus decreasing the models size.

Parameters:

index -