Module es.upm.fi.cig.multictbnc

Package es.upm.fi.cig.multictbnc.fss

Class ConInd

java.lang.Object

es.upm.fi.cig.multictbnc.fss.StatisticalBasedFeatureSelection

es.upm.fi.cig.multictbnc.fss.ConInd

All Implemented Interfaces:

OnlineFeatureSubsetSelection

public class ConInd
extends StatisticalBasedFeatureSelection
implements OnlineFeatureSubsetSelection

This class implements the ConInd online feature subset selection algorithm from Yu et al. 2018.

Constructor Summary

Constructors

Constructor	Description
ConInd(List<String> nameClassVariables, ParameterLearningAlgorithm cimPLA, int maxSeparatingSizeRedundancyAnalysis, double sigTimeTransitionHyp, double sigStateToStateTransitionHyp)	Constructs a ConInd object.

Method Summary

Modifier and Type	Method	Description
SubsetSelectedFeatures	execute(String newVariable, Dataset dataBatch)	Executes the feature subset selection algorithm for a newly arrived feature variable in a given data batch.
protected List<CIMNode>	featureSubsetSelectionGivenClassVariable(String nameNewFeatureVariable, Dataset dataBatch)	Performs feature subset selection given a new feature variable.
boolean	getLastExecutionYieldAnyChange()	Returns whether the last execution of the feature subset selection algorithm resulted in any changes.
void	setCurrentFeatureVariables(List<String> nameFeatureVariables)	Sets the current feature variables for the algorithm.

Methods inherited from class es.upm.fi.cig.multictbnc.fss.StatisticalBasedFeatureSelection

conditionalIndependenceTest, redundancyAnalysis, redundancyAnalysis, redundancyAnalysis, redundancyAnalysis, redundancyAnalysis, setDataset, testNullStateToStateTransitionHypothesis, testNullTimeToTransitionHypothesis

Methods inherited from class java.lang.Object

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

Constructor Details

ConInd

public ConInd(List<String> nameClassVariables,
 ParameterLearningAlgorithm cimPLA,
 int maxSeparatingSizeRedundancyAnalysis,
 double sigTimeTransitionHyp,
 double sigStateToStateTransitionHyp)

Constructs a ConInd object.

Parameters:

nameClassVariables - list of names of class variables

cimPLA - parameter learning algorithm for CIM nodes

maxSeparatingSizeRedundancyAnalysis - maximum size of separating sets in redundancy analysis

sigTimeTransitionHyp - significance level for time transition hypothesis tests

sigStateToStateTransitionHyp - significance level for state-to-state transition hypothesis tests

Method Details

execute

public SubsetSelectedFeatures execute(String newVariable,
 Dataset dataBatch)

Description copied from interface: OnlineFeatureSubsetSelection

Executes the feature subset selection algorithm for a newly arrived feature variable in a given data batch. This method is designed to analyze the relevance and redundancy of a new variable in the context of the current feature subset and update it accordingly.

Specified by:

execute in interface OnlineFeatureSubsetSelection

Parameters:

newVariable - name of the new feature variable that is to be evaluated

dataBatch - data batch containing the new variable along with existing features and class variables

Returns:

an instance of SubsetSelectedFeatureVariables, containing the subset of selected features

getLastExecutionYieldAnyChange

public boolean getLastExecutionYieldAnyChange()

Returns whether the last execution of the feature subset selection algorithm resulted in any changes.

Specified by:

getLastExecutionYieldAnyChange in interface OnlineFeatureSubsetSelection

Returns:

true if the last execution yielded a change, false otherwise

setCurrentFeatureVariables

public void setCurrentFeatureVariables(List<String> nameFeatureVariables)

Sets the current feature variables for the algorithm.

Specified by:

setCurrentFeatureVariables in interface OnlineFeatureSubsetSelection

Parameters:

nameFeatureVariables - list of names of current feature variables

featureSubsetSelectionGivenClassVariable

protected List<CIMNode> featureSubsetSelectionGivenClassVariable(String nameNewFeatureVariable,
 Dataset dataBatch)

Performs feature subset selection given a new feature variable.

Parameters:

nameNewFeatureVariable - the name of the new feature variable

dataBatch - the dataset on which the selection is based

Returns:

a list of CIMNode objects representing the selected feature nodes