keel.Algorithms.Decision_Trees.FunctionalTrees

Class FunctionalTrees

java.lang.Object

keel.Algorithms.Decision_Trees.FunctionalTrees.FunctionalTrees

public class FunctionalTrees
extends java.lang.Object

The Functional Trees algorithm builds a decision tree model integrating in only one model a decision tree and another classifier. The current version is called FT-Leaves because the classifier is only present at the leaf nodes. The params that can be used with this algorithm are various: - minNumInstancesToSplit, the minimum number of instances that a node should have to be considered for split - splitCriteria, which is the criteria used to decide which is the best split for a node. We consider criterias such as entropy, information gain, gini index or gain ratio - pruneCriteria, which is the criteria used to prune the tree when it has been built. The criteria used depends on a general error on the tree or a prune for all the leaves - classifierOnLeaves, the classifier that is on the leaves of the tree. The current version only supports Naive Bayes, KNN, Nearest Means, KSNN and KNN Adaptive - K, the parameter for some of the classifiers on the leaves

Since:

JDK1.5

Version:

0.1

Author:

Written by Victoria Lopez Morales (University of Granada) 24/05/2009

Constructor Summary

Constructors

Constructor and Description
FunctionalTrees(java.lang.String script) Creates a FunctionalTrees instance by reading the script file that contains all the information needed for running the algorithm

Method Summary

Modifier and Type	Method and Description
int	evaluateItem(double[] item) Classifies a given item with the information stored in the tree
void	execute() This method performs the classification for all the instances: the train and the test sets
void	growTree(TreeNode node) Builds the tree from a tree node that functions as a root node, with all the data stored in the class
void	print(myDataset data, java.lang.String filename, int type) Prints in a file the result of the classification made with the tree generated by the PUBLIC algorithm.
void	printResults(myDataset data, java.lang.String filename) Prints in a file the result of the classification made with the tree generated by the PUBLIC algorithm, this means, the tree itself and the general information about it
void	pruneTree() Prunes the tree accordingly to the prune criteria, this means, makes some of the non-leaf nodes as leaves and deletes its descendants
protected void	readConfiguration(java.lang.String script) Reads configuration script, and extracts its contents.
protected void	readParameters(java.lang.String script) Reads configuration script, to extract the parameter's values.

Methods inherited from class java.lang.Object

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

Constructor Detail

FunctionalTrees

public FunctionalTrees(java.lang.String script)

Creates a FunctionalTrees instance by reading the script file that contains all the information needed for running the algorithm

Parameters:

script - The configuration script which contains the parameters of the algorithm

Method Detail

growTree

public void growTree(TreeNode node)

Builds the tree from a tree node that functions as a root node, with all the data stored in the class

Parameters:

node - Tree node that is considered as a root node from which we are generating descendant nodes

pruneTree

public void pruneTree()

Prunes the tree accordingly to the prune criteria, this means, makes some of the non-leaf nodes as leaves and deletes its descendants

execute

public void execute()

This method performs the classification for all the instances: the train and the test sets

readConfiguration

protected void readConfiguration(java.lang.String script)

Reads configuration script, and extracts its contents.

Parameters:

script - Name of the configuration script

readParameters

protected void readParameters(java.lang.String script)

Reads configuration script, to extract the parameter's values.

Parameters:

script - Name of the configuration script

evaluateItem

public int evaluateItem(double[] item)

Classifies a given item with the information stored in the tree

Parameters:

item - Data attribute values for the item we are classifying

Returns:

the class asigned to the item given

print

public void print(myDataset data,
                  java.lang.String filename,
                  int type)

Prints in a file the result of the classification made with the tree generated by the PUBLIC algorithm. This can be done over the train set or the test set.

Parameters:

data - Dataset that we are classifying

filename - Name of the file that is going to store the results

type - 0 if we are working with a train set, 1 if we are working with a test set

printResults

public void printResults(myDataset data,
                         java.lang.String filename)

Prints in a file the result of the classification made with the tree generated by the PUBLIC algorithm, this means, the tree itself and the general information about it

Parameters:

data - Dataset that we are working with

filename - Name of the file that is going to store the results