keel.Algorithms.Hyperrectangles.INNER

Class Rule

java.lang.Object

keel.Algorithms.Hyperrectangles.INNER.Rule

All Implemented Interfaces:

java.util.Comparator

public class Rule
extends java.lang.Object
implements java.util.Comparator

File: Rule.java Auxiliary class to repressent rules for the INNER algorithm

Since:

JDK1.5

Version:

1.1

Author:

Written by Joaquin Derrac (University of Granada) 17/10/2009

Constructor Summary

Constructors

Constructor and Description
Rule() Default builder.
Rule(double[] instance, int out) Builder.

Method Summary

Modifier and Type	Method and Description
Rule	clone() Clone method
int	compare(java.lang.Object o, java.lang.Object o2) Compare two rules, regarding its impurity level
static void	computeDefaultImpurityLevels() Computes the initial confidence intervals of the impurity levels of each class
void	computeImpurityLevel() Computes the impurity level of a rule
boolean	contains(Rule another) Test the rule contains a second rule
static void	copyData(double[][] trainData, int[] trainOutput) Copies the training data iniside the class
double	distance(double[] example) Computes the distance between a rule and an instance
boolean	equals(java.lang.Object rul) Equals method
void	generalizeContinuous(int att, double value, double percentage, int out) Generalizes a continuous attribute
void	generalizeNominal(int att, double value, double percentage, int out) Generalizes a nominal attribute
double	getImpurityLevel() Returns the impurity level of a rule
int	getOutput() Returns the output class of the rule
double	getSpecialImpurityLevel(Rule[] ruleset) Computes the impurity level of a rule, by considering only those instances which are not already covered by other rules
double	inclusionDegree(Rule another, int att) Computes the inclussion degree of two rules in a given attribute
boolean	inside(double[] example) Tests if an instance is covered by the rule
boolean	intersect(Rule another) Test if two rules intersects
void	pruneConditions() Prune redundant conditions of a rule, if it decreases its impurity level
double	ruleDistance(Rule another) Computes the distance between two rules
static void	setAttributes(Attribute[] inputs) Test which attributes are nominal
static void	setNClasses(int value) Sets the number of classes of the problem
static void	setNInstances(int[] vector) Sets the array of instances of each class
static void	setNumValue(int value, int pos) Sets the number of different values for an attribute
static void	setSize(int value) Sets the size of the problem
Rule	testExtension(Rule another) Performs an extension of a rule to another
java.lang.String	toString() To String method

Methods inherited from class java.lang.Object

finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface java.util.Comparator

comparing, comparing, comparingDouble, comparingInt, comparingLong, naturalOrder, nullsFirst, nullsLast, reversed, reverseOrder, thenComparing, thenComparing, thenComparing, thenComparingDouble, thenComparingInt, thenComparingLong

Constructor Detail

Rule

public Rule()

Default builder. Generates a void rule

Rule

public Rule(double[] instance,
            int out)

Builder. Generates a rule covering only a point

Parameters:

instance - Basic instance

out - Ouput of the instance

Method Detail

setSize

public static void setSize(int value)

Sets the size of the problem

Parameters:

value - Number of attributes of the rule

setAttributes

public static void setAttributes(Attribute[] inputs)

Test which attributes are nominal

Parameters:

inputs - Attributes' descriptions

setNumValue

public static void setNumValue(int value,
                               int pos)

Sets the number of different values for an attribute

Parameters:

value - Number of values

pos - Index of the attribute

setNClasses

public static void setNClasses(int value)

Sets the number of classes of the problem

Parameters:

value - Number of classes

getOutput

public int getOutput()

Returns the output class of the rule

Returns:

Output class of the rule

copyData

public static void copyData(double[][] trainData,
                            int[] trainOutput)

Copies the training data iniside the class

Parameters:

trainData - training data Training data

trainOutput - training output Output values

setNInstances

public static void setNInstances(int[] vector)

Sets the array of instances of each class

Parameters:

vector - Array of instances of each class

computeDefaultImpurityLevels

public static void computeDefaultImpurityLevels()

Computes the initial confidence intervals of the impurity levels of each class

ruleDistance

public double ruleDistance(Rule another)

Computes the distance between two rules

Parameters:

another - Second rule

Returns:

Distance between the two rules

distance

public double distance(double[] example)

Computes the distance between a rule and an instance

Parameters:

example - The instance

Returns:

Distance between the rule and the instance

generalizeContinuous

public void generalizeContinuous(int att,
                                 double value,
                                 double percentage,
                                 int out)

Generalizes a continuous attribute

Parameters:

att - Attribute index

value - Value to obtain

percentage - Percentage of generalization allowed

out - Output class of the example to be covered

generalizeNominal

public void generalizeNominal(int att,
                              double value,
                              double percentage,
                              int out)

Generalizes a nominal attribute

Parameters:

att - Attribute index

value - Value to obtain

percentage - Percentage of generalization allowed

out - Output class of the example to be covered

computeImpurityLevel

public void computeImpurityLevel()

Computes the impurity level of a rule

getSpecialImpurityLevel

public double getSpecialImpurityLevel(Rule[] ruleset)

Computes the impurity level of a rule, by considering only those instances which are not already covered by other rules

Parameters:

ruleset - other rules to cosider.

Returns:

Impurity level

inside

public boolean inside(double[] example)

Tests if an instance is covered by the rule

Parameters:

example - Instance to be tested

Returns:

True if it is covered. False, if not.

getImpurityLevel

public double getImpurityLevel()

Returns the impurity level of a rule

Returns:

the impurity level of a rule

clone

public Rule clone()

Clone method

Overrides:

clone in class java.lang.Object

Returns:

A intialized copy of the rule

pruneConditions

public void pruneConditions()

Prune redundant conditions of a rule, if it decreases its impurity level

intersect

public boolean intersect(Rule another)

Test if two rules intersects

Parameters:

another - Second rule.

Returns:

True if the rules intersect. False, if not.

contains

public boolean contains(Rule another)

Test the rule contains a second rule

Parameters:

another - Second rule.

Returns:

True if the second rule is contained. False, if not.

inclusionDegree

public double inclusionDegree(Rule another,
                              int att)

Computes the inclussion degree of two rules in a given attribute

Parameters:

another - Second rule.

att - Attribute selected

Returns:

Degree of inclussion

testExtension

public Rule testExtension(Rule another)

Performs an extension of a rule to another

Parameters:

another - Second rule.

Returns:

Final Rule

equals

public boolean equals(java.lang.Object rul)

Equals method

Specified by:

equals in interface java.util.Comparator

Overrides:

equals in class java.lang.Object

Parameters:

rul - Another rule

Returns:

True of both rules are equal. False, if not

toString

public java.lang.String toString()

To String method

Overrides:

toString in class java.lang.Object

Returns:

A text string representing the contents of the rule

compare

public int compare(java.lang.Object o,
                   java.lang.Object o2)

Compare two rules, regarding its impurity level

Specified by:

compare in interface java.util.Comparator

Parameters:

o - First rule.

o2 - Second rule.

Returns:

Order of the rules