keel.Algorithms.Associative_Classification.ClassifierCPAR

Class PRM

java.lang.Object

keel.Algorithms.Associative_Classification.ClassifierCPAR.PRM

public class PRM
extends java.lang.Object

This class stores information to manage the PRM procedure. This algorithm modifies FOIL to achieve higher accuracy and efficiency.

Since:

JDK1.5

Version:

1.0

Author:

Written by Jesus Alcala (University of Granada) 09/02/2010

Constructor Summary

Constructors

Constructor and Description
PRM() Default Constructor
PRM(DataBase dataBase, myDataset train, RuleBase ruleBase, double alfa, double delta, double min_gain) Parameters Constructor

Method Summary

Modifier and Type	Method and Description
void	calculateAc(Rule r) It computes the PNArray A' from a rule r
void	calculatePNc(Rule r) It computes positive and negative values for each example in the training set by means a given rule
void	changePNAc(Literal lit) It modifies positive and negative values and the PNArray of each example from a given literal
void	copyA() Function to copy a PNArray to an auxiliar PNArray
void	copyPN() Function to copy P and N values to an auxiliar structure
void	generatePR() Main function of the method: it prunes the rulebase to obtain the final set of rules.
double	getTotalWeight() It returns the total weight of the positive classification
void	iniPN(int clas) It initializes positive and negative values for each example for a given class

Methods inherited from class java.lang.Object

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

Constructor Detail

PRM

public PRM()

Default Constructor

PRM

public PRM(DataBase dataBase,
           myDataset train,
           RuleBase ruleBase,
           double alfa,
           double delta,
           double min_gain)

Parameters Constructor

Parameters:

dataBase - DataBase Set of training data which is necessary to compute PRM

train - myDataset Training data set with information to compute PRM (mainly, the training examples)

ruleBase - RuleBase Set of rules to be pruned

alfa - double Weight decay factor

delta - double Number of best rules we will take in account for each example

min_gain - double Minimum gain threshold to be used

Method Detail

generatePR

public void generatePR()

Main function of the method: it prunes the rulebase to obtain the final set of rules.

iniPN

public void iniPN(int clas)

It initializes positive and negative values for each example for a given class

Parameters:

clas - int Class to initialize positive and negative values of each example

copyPN

public void copyPN()

Function to copy P and N values to an auxiliar structure

copyA

public void copyA()

Function to copy a PNArray to an auxiliar PNArray

calculatePNc

public void calculatePNc(Rule r)

It computes positive and negative values for each example in the training set by means a given rule

Parameters:

r - Rule Rule used to compute P and N

calculateAc

public void calculateAc(Rule r)

It computes the PNArray A' from a rule r

Parameters:

r - Rule Rule used to compute the PNArray

changePNAc

public void changePNAc(Literal lit)

It modifies positive and negative values and the PNArray of each example from a given literal

Parameters:

lit - Literal Given literal to modify P, N and Ac

getTotalWeight

public double getTotalWeight()

It returns the total weight of the positive classification

Returns:

double Total weight