weka.classifiers.trees.j48

Class C45ModelSelection

java.lang.Object

weka.classifiers.trees.j48.ModelSelection

weka.classifiers.trees.j48.C45ModelSelection

All Implemented Interfaces:

java.io.Serializable, RevisionHandler

public class C45ModelSelection
extends ModelSelection

Class for selecting a C4.5-type split for a given dataset.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected Instances	m_allData All the training data
protected boolean	m_doNotMakeSplitPointActualValue Do not relocate split point to actual data value
protected int	m_minNoObj Minimum number of objects in interval.
protected boolean	m_useMDLcorrection Use MDL correction?

Constructor Summary

Constructors

Constructor and Description
C45ModelSelection(int minNoObj, Instances allData, boolean useMDLcorrection, boolean doNotMakeSplitPointActualValue) Initializes the split selection method with the given parameters.

Method Summary

Modifier and Type	Method and Description
void	cleanup() Sets reference to training data to null.
java.lang.String	getRevision() Returns the revision string.
ClassifierSplitModel	selectModel(Instances data) Selects C4.5-type split for the given dataset.
ClassifierSplitModel	selectModel(Instances train, Instances test) Selects C4.5-type split for the given dataset.

Methods inherited from class java.lang.Object

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

Field Detail

m_minNoObj

protected final int m_minNoObj

Minimum number of objects in interval.

m_useMDLcorrection

protected final boolean m_useMDLcorrection

Use MDL correction?

m_allData

protected Instances m_allData

All the training data

m_doNotMakeSplitPointActualValue

protected final boolean m_doNotMakeSplitPointActualValue

Do not relocate split point to actual data value

Constructor Detail

C45ModelSelection

public C45ModelSelection(int minNoObj,
                         Instances allData,
                         boolean useMDLcorrection,
                         boolean doNotMakeSplitPointActualValue)

Initializes the split selection method with the given parameters.

Parameters:

minNoObj - minimum number of instances that have to occur in at least two subsets induced by split

allData - FULL training dataset (necessary for selection of split points).

useMDLcorrection - whether to use MDL adjustement when finding splits on numeric attributes

doNotMakeSplitPointActualValue - if true, split point is not relocated by scanning the entire dataset for the closest data value

Method Detail

cleanup

public void cleanup()

Sets reference to training data to null.

selectModel

public final ClassifierSplitModel selectModel(Instances data)
                                       throws java.lang.InterruptedException

Selects C4.5-type split for the given dataset.

Specified by:

selectModel in class ModelSelection

Throws:

java.lang.InterruptedException

selectModel

public final ClassifierSplitModel selectModel(Instances train,
                                              Instances test)
                                       throws java.lang.InterruptedException

Selects C4.5-type split for the given dataset.

Overrides:

selectModel in class ModelSelection

Throws:

java.lang.InterruptedException

getRevision

public java.lang.String getRevision()

Returns the revision string.

Returns:

the revision