weka.classifiers.rules.part
Class PruneableDecList

java.lang.Object
  |
  +--weka.classifiers.rules.part.ClassifierDecList
        |
        +--weka.classifiers.rules.part.PruneableDecList
All Implemented Interfaces:
java.io.Serializable

public class PruneableDecList
extends ClassifierDecList

Class for handling a partial tree structure that can be pruned using a pruning set.

Version:
$Revision: 1.6 $
Author:
Eibe Frank (eibe@cs.waikato.ac.nz)
See Also:
Serialized Form

Constructor Summary
PruneableDecList(ModelSelection toSelectLocModel, int minNum)
          Constructor for pruneable partial tree structure.
 
Method Summary
 void buildDecList(Instances train, Instances test, boolean leaf)
          Builds the partial tree with hold out set
 void buildRule(Instances train, Instances test)
          Method for building a pruned partial tree.
 
Methods inherited from class weka.classifiers.rules.part.ClassifierDecList
buildDecList, buildRule, chooseIndex, chooseLastIndex, classifyInstance, cleanup, distributionForInstance, toString, weight
 
Methods inherited from class java.lang.Object
equals, getClass, hashCode, notify, notifyAll, wait, wait, wait
 

Constructor Detail

PruneableDecList

public PruneableDecList(ModelSelection toSelectLocModel,
                        int minNum)
Constructor for pruneable partial tree structure.

Parameters:
toSelectLocModel - selection method for local splitting model
minNum - minimum number of objects in leaf
Method Detail

buildRule

public void buildRule(Instances train,
                      Instances test)
               throws java.lang.Exception
Method for building a pruned partial tree.

Throws:
java.lang.Exception - if tree can't be built successfully

buildDecList

public void buildDecList(Instances train,
                         Instances test,
                         boolean leaf)
                  throws java.lang.Exception
Builds the partial tree with hold out set

Throws:
java.lang.Exception - if something goes wrong


Copyright (c) 2003 David Lindsay, Computer Learning Research Centre, Dept. Computer Science, Royal Holloway, University of London