keel.Algorithms.ImbalancedClassification.ImbalancedAlgorithms.GP_COACH_H

Class Rule

java.lang.Object

keel.Algorithms.ImbalancedClassification.ImbalancedAlgorithms.GP_COACH_H.Rule

All Implemented Interfaces:

java.lang.Comparable

public class Rule
extends java.lang.Object
implements java.lang.Comparable

Title: Rule

Description: Fuzzy Rule in the GP-COACH algorithm

Company: KEEL

Since:

JDK1.5

Version:

1.5

Author:

Written by Victoria Lopez (University of Granada) 19/11/2011

Constructor Summary

Constructors

Constructor and Description
Rule() Default constructor
Rule(DataBase database, int n_classes, int tnorm, int tconorm, int rule_weight) Constructor with parameters.
Rule(DataBase database, int n_classes, int tnorm, int tconorm, int rule_weight, double[] sample, int sample_class) Constructor with parameters.
Rule(Rule original) Copy constructor for a Fuzzy rule from another Fuzzy Rule

Method Summary

Modifier and Type	Method and Description
void	addLabel(int variable_mutated, DataBase data, int low_granularity) Adds a label to the fuzzy antecedent of the given variable
void	addVar(DataBase data, int low_granularity) Adds a new variable to the fuzzy antecedent set of this rule
void	changeLabel(int variable_mutated, DataBase data, int low_granularity) Changes a label in the fuzzy antecedent of the given variable from an existing value to a non-existing value
void	clearAntecedent() Deletes all the data stored in this rule antecedent
int	compareTo(java.lang.Object a) Compares this object with the specified object for order, according to the raw_fitness measure
double	compatibility(double[] sample, boolean[] missing) Computes the compatibility degree of the rule with an input example.
double	computeLocalHierarchicalMeasure(myDataset dataset, double min_matching_degree) Computes the local hierarchical measure for this rule, according to the matching degree given for samples
boolean	containsEqualAntecedents(Rule r) Compares this rule with the specified rule to check if the rules have the same antecedents
void	deleteLabel(int variable_mutated, int low_granularity) Deletes a label to the fuzzy antecedent of the given variable
void	deleteVar(int low_granularity) Deletes a variable from the fuzzy antecedent set of this rule
boolean	equals(java.lang.Object o) Compares this rule with another rule to check if both rules are equal
void	evaluate(myDataset dataset, double alpha) Evaluates this rule, computing the raw_fitness of the rule and the weight.
void	exchangeAntecedentLabel(int pos_att_to_change, FuzzyAntecedent var, DataBase data, int low_granularity) Changes the antecedent labels for one condition in the antecedent to another labels in the antecedent
void	exchangeVariables(boolean[] current_variables, java.util.ArrayList<FuzzyAntecedent> new_variables, int low_granularity) Changes the antecedent labels for one condition in the antecedent to another labels in the antecedent
int	getClas() Obtains the class associated to this rule, consequent of the fuzzy rule
double	getFitness() Obtains the fitness associated to this rule, its raw_fitness measure
double	getGranularity() Obtains the granularity of this rule as the maximum number of labels that a fuzzy antecedent can have
int	getLevel() Obtains the level of this rule; 1 for general rules, 2 for specific rules
double	getLocalHierarchicalMeasure() Obtains the local hierarchical measure associated to this rule, computed with the compute hierarchical measures procedure
int	getnCond() Obtains the number of different conditions used in this rule antecedent
int	getnVar() Obtains the number of variables used in this rule antecedent
double	getPenalizedFitness() Obtains the penalized fitness associated to this rule, computed with the token competition procedure
int	getSeizedSamples() Computes the number of training samples that match this rule.
FuzzyAntecedent	getVar(int i) Obtains the ith variable used in this rule antecedent
double	getWeight() Obtains the weight associated to this rule
int	hashCode() Computes the hash code associated to the current rule
int	ideal() Computes the number of training samples that match this rule and are correctly classified.
boolean	isSeized(int idSample) Checks if a determined value of the training set is seized by this rule or not
boolean	not_eval() Checks if the current rule has been evaluated or not.
java.lang.String	printString(java.lang.String[] names, java.lang.String[] classes) String representation of a Fuzzy Rule in the GP-COACH-H algorithm.
void	setClass(int new_class, int low_granularity) Changes the class of the rule to a new specified class.
void	setLocalHierarchicalMeasure(double measure) Sets the local hierarchical measure field to a specified given value, corresponding to this rule local hierarchical measure according to the data set and the other rules considered
void	setPenalizedFitness(double fitness) Sets the penalized fitness field to a specified given value, corresponding to this rule penalized fitness according to the data set and the other rules considered
void	updateFuzzyLabels(double[][] granularity_tuning, myDataset dataset, double alpha) Updates the current rule according to a specified lateral tuning of the associated fuzzy labels

Methods inherited from class java.lang.Object

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

Constructor Detail

Rule

public Rule()

Default constructor

Rule

public Rule(DataBase database,
            int n_classes,
            int tnorm,
            int tconorm,
            int rule_weight)

Constructor with parameters. It creates a random general rule

Parameters:

database - Data Base associated to the general rule base that includes this rule

n_classes - Number of classes in the training set

tnorm - T-norm used to compute the compatibility degree

tconorm - T-conorm used to compute the compatibility degree

rule_weight - Rule weight of the created rule

Rule

public Rule(DataBase database,
            int n_classes,
            int tnorm,
            int tconorm,
            int rule_weight,
            double[] sample,
            int sample_class)

Constructor with parameters. It creates a specific rule from a given sample

Parameters:

database - Data Base associated to the general rule base that includes this rule

n_classes - Number of classes in the training set

tnorm - T-norm used to compute the compatibility degree

tconorm - T-conorm used to compute the compatibility degree

rule_weight - Rule weight of the created rule

sample - Data sample used to generate this rule

sample_class - Output class associated to the data sample used to generate this rule

Rule

public Rule(Rule original)

Copy constructor for a Fuzzy rule from another Fuzzy Rule

Parameters:

original - Rule which will be used as base to create another Rule

Method Detail

not_eval

public boolean not_eval()

Checks if the current rule has been evaluated or not.

Returns:

true, if this rule was evaluated previously; false, if it has never been evaluated

evaluate

public void evaluate(myDataset dataset,
                     double alpha)

Evaluates this rule, computing the raw_fitness of the rule and the weight. It also initializes the token structure for the token competition computation

Parameters:

dataset - Training dataset used in this algorithm

alpha - Alpha of the raw_fitness evaluation function

compatibility

public double compatibility(double[] sample,
                            boolean[] missing)

Computes the compatibility degree of the rule with an input example. If it has missing values, the only matching possible for the rule happens when the missing value has no condition

Parameters:

sample - The input example

missing - An array with the missing values for this input example

Returns:

the degree of compatibility

getWeight

public double getWeight()

Obtains the weight associated to this rule

Returns:

the weight associated to the fuzzy rule

getClas

public int getClas()

Obtains the class associated to this rule, consequent of the fuzzy rule

Returns:

class associated to this rule, consequent of the fuzzy rule

getLevel

public int getLevel()

Obtains the level of this rule; 1 for general rules, 2 for specific rules

Returns:

class associated to this rule, consequent of the fuzzy rule

getnVar

public int getnVar()

Obtains the number of variables used in this rule antecedent

Returns:

number of variables used in this rule antecedent

getnCond

public int getnCond()

Obtains the number of different conditions used in this rule antecedent

Returns:

number of different conditions used in this rule antecedent

getFitness

public double getFitness()

Obtains the fitness associated to this rule, its raw_fitness measure

Returns:

the fitness associated to this rule

setClass

public void setClass(int new_class,
                     int low_granularity)

Changes the class of the rule to a new specified class. It also changes the internal values that correspond to the new class value

Parameters:

new_class - class that is going to be assigned to this rule

low_granularity - number of labels that indicate low granularity for a rule

getVar

public FuzzyAntecedent getVar(int i)

Obtains the ith variable used in this rule antecedent

Parameters:

i - variable position we want to extract from the antecedent

Returns:

ith variable in this rule antecedent

exchangeAntecedentLabel

public void exchangeAntecedentLabel(int pos_att_to_change,
                                    FuzzyAntecedent var,
                                    DataBase data,
                                    int low_granularity)

Changes the antecedent labels for one condition in the antecedent to another labels in the antecedent

Parameters:

pos_att_to_change - position in this rule antecedent of the attribute that is going to change labels

var - fuzzy antecedent containing the new label for the selected attribute

data - Data Base associated to the general rule base that includes this rule

low_granularity - number of labels that indicate low granularity for a rule

exchangeVariables

public void exchangeVariables(boolean[] current_variables,
                              java.util.ArrayList<FuzzyAntecedent> new_variables,
                              int low_granularity)

Changes the antecedent labels for one condition in the antecedent to another labels in the antecedent

Parameters:

current_variables - boolean array with the selected variables from this rule

new_variables - ArrayList of FuzzyAntecedent that contains the new variables for this rule

low_granularity - number of labels that indicate low granularity for a rule

addLabel

public void addLabel(int variable_mutated,
                     DataBase data,
                     int low_granularity)

Adds a label to the fuzzy antecedent of the given variable

Parameters:

variable_mutated - position of the variable that is going to have a label added

data - Data Base associated to the general rule base that includes this rule

low_granularity - number of labels that indicate low granularity for a rule

deleteLabel

public void deleteLabel(int variable_mutated,
                        int low_granularity)

Deletes a label to the fuzzy antecedent of the given variable

Parameters:

variable_mutated - position of the variable that is going to have a label deleted

low_granularity - number of labels that indicate low granularity for a rule

changeLabel

public void changeLabel(int variable_mutated,
                        DataBase data,
                        int low_granularity)

Changes a label in the fuzzy antecedent of the given variable from an existing value to a non-existing value

Parameters:

variable_mutated - position of the variable that is going to have a label deleted

data - Data Base associated to the general rule base that includes this rule

low_granularity - number of labels that indicate low granularity for a rule

clearAntecedent

public void clearAntecedent()

Deletes all the data stored in this rule antecedent

addVar

public void addVar(DataBase data,
                   int low_granularity)

Adds a new variable to the fuzzy antecedent set of this rule

Parameters:

data - Data Base associated to the general rule base that includes this rule

low_granularity - number of labels that indicate low granularity for a rule

deleteVar

public void deleteVar(int low_granularity)

Deletes a variable from the fuzzy antecedent set of this rule

Parameters:

low_granularity - number of labels that indicate low granularity for a rule

ideal

public int ideal()

Computes the number of training samples that match this rule and are correctly classified. The rule must have been previously evaluated

Returns:

the number of training samples this rule can seize and are correctly classified

getSeizedSamples

public int getSeizedSamples()

Computes the number of training samples that match this rule. The rule must have been previously evaluated

Returns:

the number of training samples this rule can seize

setPenalizedFitness

public void setPenalizedFitness(double fitness)

Sets the penalized fitness field to a specified given value, corresponding to this rule penalized fitness according to the data set and the other rules considered

Parameters:

fitness - penalized fitness value associated to this rule

isSeized

public boolean isSeized(int idSample)

Checks if a determined value of the training set is seized by this rule or not

Parameters:

idSample - Position in the training set of the value that we want to seize with this rule

Returns:

true, if the sample is seized by this rule; false, if it is not seized

getPenalizedFitness

public double getPenalizedFitness()

Obtains the penalized fitness associated to this rule, computed with the token competition procedure

Returns:

the penalized fitness associated to this rule

getGranularity

public double getGranularity()

Obtains the granularity of this rule as the maximum number of labels that a fuzzy antecedent can have

Returns:

the granularity of this rule, maximum number of labels that a fuzzy antecedent can have

setLocalHierarchicalMeasure

public void setLocalHierarchicalMeasure(double measure)

Sets the local hierarchical measure field to a specified given value, corresponding to this rule local hierarchical measure according to the data set and the other rules considered

Parameters:

measure - local hierarchical measure associated to this rule

getLocalHierarchicalMeasure

public double getLocalHierarchicalMeasure()

Obtains the local hierarchical measure associated to this rule, computed with the compute hierarchical measures procedure

Returns:

the local hierarchical measure associated to this rule

computeLocalHierarchicalMeasure

public double computeLocalHierarchicalMeasure(myDataset dataset,
                                              double min_matching_degree)

Computes the local hierarchical measure for this rule, according to the matching degree given for samples

Parameters:

dataset - Training dataset used in this algorithm

min_matching_degree - Minimum compatibility degree that the rule must have with the sample to be considered as a matching sample

Returns:

local hierarchical measure for this rule according to the matching degree given for samples

containsEqualAntecedents

public boolean containsEqualAntecedents(Rule r)

Compares this rule with the specified rule to check if the rules have the same antecedents

Parameters:

r - Rule that is going to be compared with the current rule

Returns:

true, if the antecedents for the two rules are the same; false, otherwise

updateFuzzyLabels

public void updateFuzzyLabels(double[][] granularity_tuning,
                              myDataset dataset,
                              double alpha)

Updates the current rule according to a specified lateral tuning of the associated fuzzy labels

Parameters:

granularity_tuning - Lateral tuning of the associated fuzzy labels of this rule

dataset - Training dataset used in this algorithm

alpha - Alpha of the raw_fitness evaluation function

compareTo

public int compareTo(java.lang.Object a)

Compares this object with the specified object for order, according to the raw_fitness measure

Specified by:

compareTo in interface java.lang.Comparable

Parameters:

a - object to compare with.

Returns:

a negative integer, zero, or a positive integer as this object is less than, equal to, or greater than the specified object

equals

public boolean equals(java.lang.Object o)

Compares this rule with another rule to check if both rules are equal

Overrides:

equals in class java.lang.Object

Parameters:

o - Rule that is going to be compared with the current rule

Returns:

true, if the rules are the same; false, otherwise

hashCode

public int hashCode()

Computes the hash code associated to the current rule

Overrides:

hashCode in class java.lang.Object

Returns:

the hash code associated to the current rule

printString

public java.lang.String printString(java.lang.String[] names,
                                    java.lang.String[] classes)

String representation of a Fuzzy Rule in the GP-COACH-H algorithm.

Parameters:

names - attributes names.

classes - classes considered.

Returns:

String an string containing the Fuzzy Rule