weka.classifiers.trees.j48
Class C45PruneableClassifierTree

java.lang.Object
  weka.classifiers.trees.j48.ClassifierTree
      weka.classifiers.trees.j48.C45PruneableClassifierTree

All Implemented Interfaces:

Drawable, java.io.Serializable

public class C45PruneableClassifierTree

extends ClassifierTree

Class for handling a tree structure that can be pruned using C4.5 procedures.

Version:

$Revision: 1.11 $

Author:

Eibe Frank (eibe@cs.waikato.ac.nz)

See Also:

Serialized Form

Field Summary

(package private) float	m_CF The confidence factor for pruning.
(package private) boolean	m_cleanup Cleanup after the tree has been built.
(package private) boolean	m_pruneTheTree True if the tree is to be pruned.
(package private) boolean	m_subtreeRaising Is subtree raising to be performed?

Fields inherited from class weka.classifiers.trees.j48.ClassifierTree

m_id, m_isEmpty, m_isLeaf, m_localModel, m_sons, m_test, m_toSelectModel, m_train

Fields inherited from interface weka.core.Drawable

BayesNet, NOT_DRAWABLE, TREE

Constructor Summary

C45PruneableClassifierTree(ModelSelection toSelectLocModel, boolean pruneTree, float cf, boolean raiseTree, boolean cleanup)
Constructor for pruneable tree structure.

Method Summary

void	buildClassifier(Instances data) Method for building a pruneable classifier tree.
void	collapse() Collapses a tree to a node if training error doesn't increase.
private double	getEstimatedErrors() Computes estimated errors for tree.
private double	getEstimatedErrorsForBranch(Instances data) Computes estimated errors for one branch.
private double	getEstimatedErrorsForDistribution(Distribution theDistribution) Computes estimated errors for leaf.
protected ClassifierTree	getNewTree(Instances data) Returns a newly created tree.
private double	getTrainingErrors() Computes errors of tree on training data.
private ClassifierSplitModel	localModel() Method just exists to make program easier to read.
private void	newDistribution(Instances data) Computes new distributions of instances for nodes in tree.
void	prune() Prunes a tree using C4.5's pruning procedure.
private C45PruneableClassifierTree	son(int index) Method just exists to make program easier to read.

Methods inherited from class weka.classifiers.trees.j48.ClassifierTree

assignIDs, buildTree, buildTree, classifyInstance, cleanup, distributionForInstance, getNewTree, graph, graphType, nextID, numLeaves, numNodes, prefix, resetID, toSource, toString

Methods inherited from class java.lang.Object

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

Field Detail

m_pruneTheTree

boolean m_pruneTheTree

True if the tree is to be pruned.

m_CF

float m_CF

The confidence factor for pruning.

m_subtreeRaising

boolean m_subtreeRaising

Is subtree raising to be performed?

m_cleanup

boolean m_cleanup

Cleanup after the tree has been built.

Constructor Detail

C45PruneableClassifierTree

public C45PruneableClassifierTree(ModelSelection toSelectLocModel,
                                  boolean pruneTree,
                                  float cf,
                                  boolean raiseTree,
                                  boolean cleanup)
                           throws java.lang.Exception

Constructor for pruneable tree structure. Stores reference to associated training data at each node.

Parameters:

toSelectLocModel - selection method for local splitting model

pruneTree - true if the tree is to be pruned

cf - the confidence factor for pruning

Throws:

java.lang.Exception - if something goes wrong

Method Detail

buildClassifier

public void buildClassifier(Instances data)
                     throws java.lang.Exception

Method for building a pruneable classifier tree.

Overrides:

buildClassifier in class ClassifierTree

Throws:

java.lang.Exception - if something goes wrong

collapse

public final void collapse()

Collapses a tree to a node if training error doesn't increase.

prune

public void prune()
           throws java.lang.Exception

Prunes a tree using C4.5's pruning procedure.

Throws:

java.lang.Exception - if something goes wrong

getNewTree

protected ClassifierTree getNewTree(Instances data)
                             throws java.lang.Exception

Returns a newly created tree.

Overrides:

getNewTree in class ClassifierTree

Parameters:

data - the training data

Throws:

java.lang.Exception - if something goes wrong

getEstimatedErrors

private double getEstimatedErrors()

Computes estimated errors for tree.

getEstimatedErrorsForBranch

private double getEstimatedErrorsForBranch(Instances data)
                                    throws java.lang.Exception

Computes estimated errors for one branch.

Throws:

java.lang.Exception - if something goes wrong

getEstimatedErrorsForDistribution

private double getEstimatedErrorsForDistribution(Distribution theDistribution)

Computes estimated errors for leaf.

getTrainingErrors

private double getTrainingErrors()

Computes errors of tree on training data.

localModel

private ClassifierSplitModel localModel()

Method just exists to make program easier to read.

newDistribution

private void newDistribution(Instances data)
                      throws java.lang.Exception

Computes new distributions of instances for nodes in tree.

Throws:

java.lang.Exception - if something goes wrong

son

private C45PruneableClassifierTree son(int index)

Method just exists to make program easier to read.