mulan.classifier.meta.thresholding
Class Meta

java.lang.Object
  mulan.classifier.MultiLabelLearnerBase
      mulan.classifier.meta.MultiLabelMetaLearner
          mulan.classifier.meta.thresholding.Meta

All Implemented Interfaces:

Serializable, MultiLabelLearner, TechnicalInformationHandler

Direct Known Subclasses:

MetaLabeler, ThresholdPrediction

public abstract class Meta
extends MultiLabelMetaLearner

Base class for instance-based prediction of a bipartition from the labels' scores

Version:

2010.12.14

Author:

Marios Ioannou, George Sakkas, Grigorios Tsoumakas

See Also:

Serialized Form

Field Summary

protected Classifier	classifier the classifier to learn the number of top labels or the threshold
protected Instances	classifierInstances the training instances for the single-label model
protected MultiLabelLearner	foldLearner clean multi-label learner for cross-validation
protected int	kFoldsCV the number of folds for cross validation
protected String	metaDatasetChoice the type for constructing the meta dataset

Fields inherited from class mulan.classifier.meta.MultiLabelMetaLearner

baseLearner

Fields inherited from class mulan.classifier.MultiLabelLearnerBase

featureIndices, labelIndices, numLabels

Constructor Summary

Meta(MultiLabelLearner baseLearner, Classifier aClassifier, String aMetaDatasetChoice)
Constructor that initializes the learner

Method Summary

protected void	buildInternal(MultiLabelInstances trainingData) Learner specific implementation of building the model from MultiLabelInstances training data set.
Classifier	getClassifier() Returns the classifier used to predict the number of labels/threshold
protected Instance	modifiedInstanceX(Instance instance, String xBased) A method that modify an instance
protected Instances	prepareClassifierInstances(MultiLabelInstances data) Prepares the instances for the predictor of labels/threshold
protected abstract Instances	transformData(MultiLabelInstances trainingData) abstract method that transforms the training data to meta data
protected void	valuesX(MultiLabelLearner learner, Instance instance, double[] newValues, String xBased) A method that fill the array "newValues"

Methods inherited from class mulan.classifier.meta.MultiLabelMetaLearner

getBaseLearner

Methods inherited from class mulan.classifier.MultiLabelLearnerBase

build, debug, getDebug, getTechnicalInformation, globalInfo, isModelInitialized, isUpdatable, makeCopy, makePrediction, makePredictionInternal, setDebug

Methods inherited from class java.lang.Object

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

Field Detail

classifier

protected Classifier classifier

the classifier to learn the number of top labels or the threshold

classifierInstances

protected Instances classifierInstances

the training instances for the single-label model

metaDatasetChoice

protected String metaDatasetChoice

the type for constructing the meta dataset

kFoldsCV

protected int kFoldsCV

the number of folds for cross validation

foldLearner

protected MultiLabelLearner foldLearner

clean multi-label learner for cross-validation

Constructor Detail

Meta

public Meta(MultiLabelLearner baseLearner,
            Classifier aClassifier,
            String aMetaDatasetChoice)

Constructor that initializes the learner

Parameters:

baseLearner - the MultiLabelLearner

aClassifier - the learner that will predict the number of relevant labels or a threshold

aMetaDatasetChoice - what features to use for predicting the number of relevant labels or a threshold

Method Detail

getClassifier

public Classifier getClassifier()

Returns the classifier used to predict the number of labels/threshold

Returns:

the classifier used to predict the number of labels/threshold

transformData

protected abstract Instances transformData(MultiLabelInstances trainingData)
                                    throws Exception

abstract method that transforms the training data to meta data

Parameters:

trainingData - the training data set

Returns:

the meta data for training the predictor of labels/threshold

Throws:

Exception

modifiedInstanceX

protected Instance modifiedInstanceX(Instance instance,
                                     String xBased)

A method that modify an instance

Parameters:

instance - to modified

xBased - the type for constructing the meta dataset

Returns:

a transformed instance for the predictor of labels/threshold

prepareClassifierInstances

protected Instances prepareClassifierInstances(MultiLabelInstances data)

Prepares the instances for the predictor of labels/threshold

Parameters:

data - the training data

Returns:

the prepared instances

valuesX

protected void valuesX(MultiLabelLearner learner,
                       Instance instance,
                       double[] newValues,
                       String xBased)
                throws Exception

A method that fill the array "newValues"

Parameters:

learner - the multi-label learner

instance - the training instances

newValues - the array to fill

xBased - the type for constructing the meta dataset

Throws:

Exception

buildInternal

protected void buildInternal(MultiLabelInstances trainingData)
                      throws Exception

Description copied from class: MultiLabelLearnerBase

Learner specific implementation of building the model from MultiLabelInstances training data set. This method is called from MultiLabelLearnerBase.build(MultiLabelInstances) method, where behavior common across all learners is applied.

Specified by:

buildInternal in class MultiLabelLearnerBase

Parameters:

trainingData - the training data set

Throws:

Exception - if learner model was not created successfully