edu.umass.cs.mallet.base.classify

Class DecisionTreeTrainer

java.lang.Object

edu.umass.cs.mallet.base.classify.ClassifierTrainer

edu.umass.cs.mallet.base.classify.DecisionTreeTrainer

All Implemented Interfaces:

Boostable

public class DecisionTreeTrainer
extends ClassifierTrainer
implements Boostable

A decision tree learner, roughly ID3. Does not yet implement splitting of continuous-valued features, but it should in the future. Currently a feature is considered "present" if it has positive value. ftp://ftp.cs.cmu.edu/project/jair/volume4/quinlan96a.ps Only set up for conventiently learning decision stubs: there is no pruning or good stopping rule. Currently only stop by reaching a maximum depth.

Author:

Andrew McCallum mccallum@cs.umass.edu

Constructor Summary

Constructors

Constructor and Description
DecisionTreeTrainer()
DecisionTreeTrainer(int maxDepth)

Method Summary

Modifier and Type	Method and Description
protected void	splitTree(DecisionTree.Node node, FeatureSelection selectedFeatures, int depth)
Classifier	train(InstanceList trainingList, InstanceList validationList, InstanceList testSet, ClassifierEvaluating evaluator, Classifier initialClassifier) Return a new classifier tuned using the three arguments.

Methods inherited from class edu.umass.cs.mallet.base.classify.ClassifierTrainer

main, toString, train, train, train, train

Methods inherited from class java.lang.Object

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

Constructor Detail

DecisionTreeTrainer

public DecisionTreeTrainer(int maxDepth)

DecisionTreeTrainer

public DecisionTreeTrainer()

Method Detail

splitTree

protected void splitTree(DecisionTree.Node node,
                         FeatureSelection selectedFeatures,
                         int depth)

train

public Classifier train(InstanceList trainingList,
                        InstanceList validationList,
                        InstanceList testSet,
                        ClassifierEvaluating evaluator,
                        Classifier initialClassifier)

Description copied from class: ClassifierTrainer

Return a new classifier tuned using the three arguments.

Specified by:

train in class ClassifierTrainer

Parameters:

trainingList - examples used to set parameters.

validationList - examples used to tune meta-parameters. May be null.

testSet - examples not examined at all for training, but passed on to diagnostic routines. May be null.

initialClassifier - training process may start from here. The parameters of the initialClassifier are not modified. May be null.