weka.classifiers.rules
Class NNge.Exemplar

java.lang.Object
  weka.core.Instances
      weka.classifiers.rules.NNge.Exemplar

All Implemented Interfaces:

java.io.Serializable

Enclosing class:

NNge

private class NNge.Exemplar

extends Instances

Implements Exemplar as used by NNge : parallel axis hyperrectangle.

Field Summary

private double	m_ClassValue class of the Exemplar
private double[]	m_MaxBorder The max borders of the rectangle for numeric attributes
private double[]	m_MinBorder The min borders of the rectangle for numeric attributes
private int	m_NegativeCount Number of incorrect prediction for this examplar
private NNge	m_NNge The NNge which owns this Exemplar
private int	m_PositiveCount Number of correct prediction for this examplar
private Instance	m_PreInst
private double[]	m_PreMaxBorder the arrays used by preGeneralise
private double[]	m_PreMinBorder
private boolean[][]	m_PreRange
private boolean[][]	m_Range The ranges of the hyperrectangle for nominal attributes
private NNge.Exemplar	next
private NNge.Exemplar	nextWithClass
private NNge.Exemplar	previous List of all the Exemplar
private NNge.Exemplar	previousWithClass List of all the Exemplar with the same class

Fields inherited from class weka.core.Instances

FILE_EXTENSION, m_Attributes, m_ClassIndex, m_IndicesBuffer, m_Instances, m_RelationName, m_ValueBuffer

Constructor Summary

private

NNge.Exemplar(NNge nnge, Instances inst, int size, double classV) Build a new empty Exemplar

Method Summary

private double	attrDistance(Instance inst, int attrIndex) Compute the distance between the projection of inst and this Exemplar along the attribute attrIndex.
private void	cancelGeneralisation() Cancels a generalisation started with preGeneralise.
private double	classValue() Return the class of the Exemplar
private void	generalise(Instance inst) Generalise the Exemplar with inst
private double	getMaxBorder(int attrIndex) Returns the value of the sup border of the hyperrectangle Returns NaN if the HyperRectangle doesn't have any border for this attribute
private double	getMinBorder(int attrIndex) Returns the value of the inf border of the Exemplar.
private int	getNegativeCount() Returns the number of negative classifications
private int	getPositiveCount() Returns the number of positive classifications
private boolean	holds(Instance inst) return true if inst is held by this Exemplar, false otherwise
private boolean	holds(int attrIndex, double value) return true if value is inside the Exemplar along the attrIndex attribute.
private void	incrNegativeCount() Increment the number of negative Classifications
private void	incrPositiveCount() Increment the number of positive Classifications
private boolean	isEmpty() Returns true if the Exemplar is empty (i.e. doesn't yield any Instance)
private boolean	overlaps(NNge.Exemplar ex) Check if the Examplar overlaps ex
private void	preGeneralise(Instance inst) pre-generalise the Exemplar with inst i.e. the boundaries of the Exemplar include inst but the Exemplar still doesn't 'own' inst.
private void	setNegativeCount(int value) Set the number of negative classifications
private void	setPositiveCount(int value) Set the number of positive classifications
private double	squaredDistance(Instance inst) Returns the square of the distance between inst and the Exemplar.
private java.lang.String	toRules() Returns a string of the rules induced by this examplar
private java.lang.String	toString2() Returns a description of this Exemplar
private void	validateGeneralisation() Validates a generalisation started with preGeneralise.
private double	weight() Return the weight of the Examplar

Methods inherited from class weka.core.Instances

add, attribute, attribute, attributeStats, attributeToDoubleArray, checkForStringAttributes, checkInstance, classAttribute, classIndex, compactify, delete, delete, deleteAttributeAt, deleteStringAttributes, deleteWithMissing, deleteWithMissing, deleteWithMissingClass, enumerateAttributes, enumerateInstances, equalHeaders, firstInstance, getInstance, getInstanceFull, getInstanceSparse, getRandomNumberGenerator, insertAttributeAt, instance, lastInstance, main, meanOrMode, meanOrMode, mergeInstances, numAttributes, numClasses, numDistinctValues, numDistinctValues, numInstances, randomize, readHeader, readInstance, relationName, renameAttribute, renameAttribute, renameAttributeValue, renameAttributeValue, resample, resampleWithWeights, resampleWithWeights, setClass, setClassIndex, setRelationName, sort, sort, stratify, stringFreeStructure, sumOfWeights, test, testCV, toString, toSummaryString, trainCV, trainCV, variance, variance

Methods inherited from class java.lang.Object

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

Field Detail

previous

private NNge.Exemplar previous

List of all the Exemplar

next

private NNge.Exemplar next

previousWithClass

private NNge.Exemplar previousWithClass

List of all the Exemplar with the same class

nextWithClass

private NNge.Exemplar nextWithClass

m_NNge

private NNge m_NNge

The NNge which owns this Exemplar

m_ClassValue

private double m_ClassValue

class of the Exemplar

m_PositiveCount

private int m_PositiveCount

Number of correct prediction for this examplar

m_NegativeCount

private int m_NegativeCount

Number of incorrect prediction for this examplar

m_MaxBorder

private double[] m_MaxBorder

The max borders of the rectangle for numeric attributes

m_MinBorder

private double[] m_MinBorder

The min borders of the rectangle for numeric attributes

m_Range

private boolean[][] m_Range

The ranges of the hyperrectangle for nominal attributes

m_PreMaxBorder

private double[] m_PreMaxBorder

the arrays used by preGeneralise

m_PreMinBorder

private double[] m_PreMinBorder

m_PreRange

private boolean[][] m_PreRange

m_PreInst

private Instance m_PreInst

Constructor Detail

NNge.Exemplar

private NNge.Exemplar(NNge nnge,
                      Instances inst,
                      int size,
                      double classV)

Build a new empty Exemplar

Parameters:

nnge - the classifier which owns this Exemplar

inst - the instances from which the header information is to be taken

size - the capacity of the Exemplar

classV - the class of the Exemplar

Method Detail

generalise

private void generalise(Instance inst)
                 throws java.lang.Exception

Generalise the Exemplar with inst

Throws:

java.lang.Exception - if either the class of inst is not equal to the class of the Exemplar or inst misses a value.

preGeneralise

private void preGeneralise(Instance inst)
                    throws java.lang.Exception

pre-generalise the Exemplar with inst i.e. the boundaries of the Exemplar include inst but the Exemplar still doesn't 'own' inst. To be complete, the generalisation must be validated with validateGeneralisation. the generalisation can be canceled with cancelGeneralisation.

Throws:

java.lang.Exception - if either the class of inst is not equal to the class of the Exemplar or inst misses a value.

validateGeneralisation

private void validateGeneralisation()
                             throws java.lang.Exception

Validates a generalisation started with preGeneralise. Watch out, preGeneralise must have been called before.

Throws:

java.lang.Exception - is thrown if preGeneralise hasn't been called before

cancelGeneralisation

private void cancelGeneralisation()
                           throws java.lang.Exception

Cancels a generalisation started with preGeneralise. Watch out, preGeneralise must have been called before.

Throws:

java.lang.Exception - is thrown if preGeneralise hasn't been called before

holds

private boolean holds(Instance inst)

return true if inst is held by this Exemplar, false otherwise

Parameters:

inst - an Instance

Returns:

true if inst is held by this hyperrectangle, false otherwise

holds

private boolean holds(int attrIndex,
                      double value)

return true if value is inside the Exemplar along the attrIndex attribute.

Parameters:

attrIndex - the index of an attribute

value - a value along the attrIndexth attribute

Returns:

true if value is inside the Exemplar along the attrIndex attribute.

overlaps

private boolean overlaps(NNge.Exemplar ex)

Check if the Examplar overlaps ex

Parameters:

ex - an Exemplar

Returns:

true if ex is overlapped by the Exemplar

Throws:

java.lang.Exception

attrDistance

private double attrDistance(Instance inst,
                            int attrIndex)

Compute the distance between the projection of inst and this Exemplar along the attribute attrIndex. If inst misses its value along the attribute, the function returns 0.

Parameters:

inst - an instance

attrIndex - the index of the attribute

Returns:

the distance between the projection of inst and this Exemplar along the attribute attrIndex.

squaredDistance

private double squaredDistance(Instance inst)

Returns the square of the distance between inst and the Exemplar.

Parameters:

inst - an instance

Returns:

the squared distance between inst and the Exemplar.

weight

private double weight()

Return the weight of the Examplar

Returns:

the weight of the Examplar.

classValue

private double classValue()

Return the class of the Exemplar

Returns:

the class of this exemplar as a double (weka format)

getMinBorder

private double getMinBorder(int attrIndex)
                     throws java.lang.Exception

Returns the value of the inf border of the Exemplar.

Parameters:

attrIndex - the index of the attribute

Returns:

the value of the inf border for this attribute

Throws:

java.lang.Exception - is thrown either if the attribute is nominal or if the Exemplar is empty

getMaxBorder

private double getMaxBorder(int attrIndex)
                     throws java.lang.Exception

Returns the value of the sup border of the hyperrectangle Returns NaN if the HyperRectangle doesn't have any border for this attribute

Parameters:

attrIndex - the index of the attribute

Returns:

the value of the sup border for this attribute

Throws:

java.lang.Exception - is thrown either if the attribute is nominal or if the Exemplar is empty

getPositiveCount

private int getPositiveCount()

Returns the number of positive classifications

Returns:

the number of positive classifications

getNegativeCount

private int getNegativeCount()

Returns the number of negative classifications

Returns:

the number of negative classifications

setPositiveCount

private void setPositiveCount(int value)

Set the number of positive classifications

Parameters:

value - an integer value (greater than 0 is wise...)

setNegativeCount

private void setNegativeCount(int value)

Set the number of negative classifications

Parameters:

value - an integer value

incrPositiveCount

private void incrPositiveCount()

Increment the number of positive Classifications

incrNegativeCount

private void incrNegativeCount()

Increment the number of negative Classifications

isEmpty

private boolean isEmpty()

Returns true if the Exemplar is empty (i.e. doesn't yield any Instance)

Returns:

true if the Exemplar is empty, false otherwise

toString2

private java.lang.String toString2()

Returns a description of this Exemplar

Returns:

A string that describes this Exemplar

toRules

private java.lang.String toRules()

Returns a string of the rules induced by this examplar

Returns:

a string of the rules induced by this examplar