keel.Algorithms.Discretizers.ExtendedChi2_Discretizer

Class Discretizer

java.lang.Object

keel.Algorithms.Discretizers.ExtendedChi2_Discretizer.Discretizer

Direct Known Subclasses:

ExtendedChi2Discretizer

public abstract class Discretizer
extends java.lang.Object

File: Discretizer.java A general framework for Chi Discretizers. This class contains all common operations in the development of a discretizer algorithm. Any discretizer can extend this class and, by implementing the abstract "discretizeAttribute" method, getting most of its work already done.

Since:

JDK1.5

Version:

1.1

Author:

Salvador García (University of Jaén - Jaén) 31/12/2009

Field Summary

Fields

Modifier and Type	Field and Description
protected int[]	classOfInstances The class of each instances.
protected double[][]	cutPoints Cut points for each attribute used to discretize them.
protected double[][]	cutPointsTMP Temporal Cut points.
protected double	inconsistencyThreshold Inconsistency Threshold.
protected boolean[]	realAttributes Boolean array identifying which attribute is continuous.
protected double[][]	realValues For each attribute, stores its values of every instances.

Constructor Summary

Constructors

Constructor and Description
Discretizer()

Method Summary

Modifier and Type	Method and Description
void	applyDiscretization(java.lang.String in, java.lang.String out) Applies the discretization stored on the cut points.
void	buildCutPoints(InstanceSet is) Creates the cut points for each attribute with the given dataset.
int	discretize(int attribute, double value) Discretizes the given value of the given attribute.
protected abstract java.util.Vector	discretizeAttribute(int attribute, int[] values, java.util.Vector<Interval> intervals, double levelSig) This abstract method creates the cut points of the attribute given using its values for each instances given.
protected abstract java.util.Vector	discretizeAttributePreliminary(int attribute, int[] values, java.util.Vector<Interval> intervals) This abstract method creates a preliminary cut points of the attribute given using its values for each instances given without sigma level.
int	discretizeTMP(int attribute, double value) Discretizes the given value of the given attribute with the temporal cut points.
double	getCutPoint(int attribute, int cp) Returns the cp-th cut point of the given attribute.
int	getNumIntervals(int attribute) Returns the number of intervals in the given attribute.
double	inconsistencyCheck(Instance[] instances) Returns the inconsistency percentage of the given instances.
protected abstract java.util.Vector<Interval>	obtainIntervals(int attribute, int[] values, int begin, int end) Returns the discretized intervals of the given attribute.
protected void	sortValues(int attribute, int[] values, int begin, int end) Sorts the instances values of the attribute given.

Methods inherited from class java.lang.Object

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

Field Detail

cutPoints

protected double[][] cutPoints

Cut points for each attribute used to discretize them.

cutPointsTMP

protected double[][] cutPointsTMP

Temporal Cut points.

realValues

protected double[][] realValues

For each attribute, stores its values of every instances.

realAttributes

protected boolean[] realAttributes

Boolean array identifying which attribute is continuous.

classOfInstances

protected int[] classOfInstances

The class of each instances.

inconsistencyThreshold

protected double inconsistencyThreshold

Inconsistency Threshold. Chi sigma parameter.

Constructor Detail

Discretizer

public Discretizer()

Method Detail

buildCutPoints

public void buildCutPoints(InstanceSet is)

Creates the cut points for each attribute with the given dataset. The cut points are generated by the execution of the abstract method "discretizeAttribute" where the different discretize will be implemented.

Parameters:

is - given dataset.

inconsistencyCheck

public double inconsistencyCheck(Instance[] instances)

Returns the inconsistency percentage of the given instances.

Parameters:

instances - given instances.

Returns:

the inconsistency percentage of the given instances.

applyDiscretization

public void applyDiscretization(java.lang.String in,
                                java.lang.String out)

Applies the discretization stored on the cut points.

Parameters:

in - original dataset filename.

out - discretized dataset filename.

sortValues

protected void sortValues(int attribute,
                          int[] values,
                          int begin,
                          int end)

Sorts the instances values of the attribute given.

Parameters:

attribute - attribute to be sorted.

values - sorted indeces.

begin - First position of the section to sort.

end - Last position of the section to sort.

getNumIntervals

public int getNumIntervals(int attribute)

Returns the number of intervals in the given attribute.

Parameters:

attribute - given attribute.

Returns:

the number of intervals in the given attribute.

getCutPoint

public double getCutPoint(int attribute,
                          int cp)

Returns the cp-th cut point of the given attribute.

Parameters:

attribute - given attribute.

cp - given cut point index.

Returns:

the cp-th cut point of the given attribute.

discretizeAttribute

protected abstract java.util.Vector discretizeAttribute(int attribute,
                                                        int[] values,
                                                        java.util.Vector<Interval> intervals,
                                                        double levelSig)

This abstract method creates the cut points of the attribute given using its values for each instances given. Every Chi discretizer should implement this method with its particular scheme.

Parameters:

attribute - given attribute to discretize.

values - given attribute values in the dataset.

intervals - given intervals.

levelSig - Sigma level.

Returns:

the cut points of the attribute given

discretizeAttributePreliminary

protected abstract java.util.Vector discretizeAttributePreliminary(int attribute,
                                                                   int[] values,
                                                                   java.util.Vector<Interval> intervals)

This abstract method creates a preliminary cut points of the attribute given using its values for each instances given without sigma level. Every Chi discretizer should implement this method with its particular scheme.

Parameters:

attribute - given attribute to discretize.

values - given attribute values in the dataset.

intervals - given intervals.

Returns:

the cut points of the attribute given

obtainIntervals

protected abstract java.util.Vector<Interval> obtainIntervals(int attribute,
                                                              int[] values,
                                                              int begin,
                                                              int end)

Returns the discretized intervals of the given attribute.

Parameters:

attribute - given attribute to discretize.

values - given attribute values in the dataset.

begin - First position of the section to discretize.

end - Last position of the section to discretize.

Returns:

the discretized intervals of the given attribute.

discretize

public int discretize(int attribute,
                      double value)

Discretizes the given value of the given attribute.

Parameters:

attribute - given attribute.

value - given value to discretize.

Returns:

the discretized value.

discretizeTMP

public int discretizeTMP(int attribute,
                         double value)

Discretizes the given value of the given attribute with the temporal cut points.

Parameters:

attribute - given attribute.

value - given value to discretize.

Returns:

the discretized value.