keel.Algorithms.Genetic_Rule_Learning.OIGA

Class RuleSet

java.lang.Object

keel.Algorithms.Genetic_Rule_Learning.OIGA.RuleSet

All Implemented Interfaces:

java.lang.Comparable

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

This class represents a set of rules in the OIGA algorithm

Since:

JDK 1.5

Version:

0.2

Author:

Written by Julián Luengo Martín 08/02/2007

Constructor Summary

Constructors

Constructor and Description
RuleSet() Default constructor.
RuleSet(int numberRules, int numAtt) Constructor for a fixed number of rules and attributes
RuleSet(RuleSet orig) Deep-copy constructor

Method Summary

Modifier and Type	Method and Description
int	classify(Instance inst) Classifies an instance using the set of rules
double	classify(InstanceSet ISet) Classifies the instances of a data sets, and updates the fitness function as the number of well classified instances
int	compareTo(java.lang.Object o) It compares the RuleSet with other one
void	copyFromBegintoPoint(RuleSet rs, int cutpoint_rule, int cutpoint_variable) Copies the rules from the beginning of the rule set to the selected cutpoint
void	copyFromPointtoEnd(RuleSet rs, int cutpoint_rule, int cutpoint_variable) Copies the rules from cutpoint to the end of the rule set
void	createRules(int numberRules, int numAtt) Reset the current rules, and creates a new -clean- set
boolean	equals(RuleSet rs) Test if the fitness of the rule sets are equal
double	getFitness() Gets the fitness of this rule set
void	iga(RuleSet rs) Incremental Genetic Algorithm, which increases the size of the actual rule set appending all the attributes from rs to the rules of the actual rule set.
boolean	isEvaluated() Test if the rule set is currently evaluated
void	mutate(int gene) Mutate a variable of the rule set (i.e. the activation, limits or class of a rule from the rule set)
void	randomizeRules(InstanceSet IS) Initialize the set of rules
void	setEvaluated(boolean eval) Returns the evaluation state

Methods inherited from class java.lang.Object

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

Constructor Detail

RuleSet

public RuleSet()

Default constructor. No memory allocation

RuleSet

public RuleSet(int numberRules,
               int numAtt)

Constructor for a fixed number of rules and attributes

Parameters:

numberRules - number of rules of the set

numAtt - number of attributes of each rule

RuleSet

public RuleSet(RuleSet orig)

Deep-copy constructor

Parameters:

orig - the set of rules which will be copied over this set

Method Detail

createRules

public void createRules(int numberRules,
                        int numAtt)

Reset the current rules, and creates a new -clean- set

Parameters:

numberRules - number of rules of the set

numAtt - number of attributes of each rule

randomizeRules

public void randomizeRules(InstanceSet IS)

Initialize the set of rules

Parameters:

IS - train data set used for initialization

classify

public double classify(InstanceSet ISet)

Classifies the instances of a data sets, and updates the fitness function as the number of well classified instances

Parameters:

ISet - the data set that will be classified

Returns:

the obtained fitness

classify

public int classify(Instance inst)

Classifies an instance using the set of rules

Parameters:

inst - the instance that will be classified

Returns:

the predicted class, or -1 if not covered by any rule (unknown class)

copyFromPointtoEnd

public void copyFromPointtoEnd(RuleSet rs,
                               int cutpoint_rule,
                               int cutpoint_variable)

Copies the rules from cutpoint to the end of the rule set

Parameters:

rs - rule set from which the rules will be copied

cutpoint_rule - the rule in which there is the cutpoint

cutpoint_variable - the variable (activation, limits or class) in which the cutpoint is

copyFromBegintoPoint

public void copyFromBegintoPoint(RuleSet rs,
                                 int cutpoint_rule,
                                 int cutpoint_variable)

Copies the rules from the beginning of the rule set to the selected cutpoint

Parameters:

rs - rule set from which the rules will be copied

cutpoint_rule - the rule in which there is the cutpoint

cutpoint_variable - the variable (activation, limits or class) in which the cutpoint is

mutate

public void mutate(int gene)

Mutate a variable of the rule set (i.e. the activation, limits or class of a rule from the rule set)

Parameters:

gene - the gene (variable) that will be mutated

iga

public void iga(RuleSet rs)

Incremental Genetic Algorithm, which increases the size of the actual rule set appending all the attributes from rs to the rules of the actual rule set. The rules used for this task should have the same class as the rule from our rule set, but if we cannot find it, we use any rule.

Parameters:

rs - the rule set from which we will append the attributes

isEvaluated

public boolean isEvaluated()

Test if the rule set is currently evaluated

Returns:

if the rule set is evaluated, so its fitness is OK

setEvaluated

public void setEvaluated(boolean eval)

Returns the evaluation state

Parameters:

eval - the evaluation state (true or false) of this rule set

getFitness

public double getFitness()

Gets the fitness of this rule set

Returns:

the current fitness of the rule set (updated or NOT!)

compareTo

public int compareTo(java.lang.Object o)

It compares the RuleSet with other one

Specified by:

compareTo in interface java.lang.Comparable

Parameters:

o - the rule set to compare

Returns:

0 if both rule sets have the same fitness, 1 if the fitness of this is higher, and -1 otherwise

equals

public boolean equals(RuleSet rs)

Test if the fitness of the rule sets are equal

Parameters:

rs - the rule set which will be compared to ours

Returns:

if they have equal fitness or not