weka.classifiers.rules
Class NNge

java.lang.Object
  weka.classifiers.Classifier
      weka.classifiers.rules.NNge

All Implemented Interfaces:

java.lang.Cloneable, OptionHandler, java.io.Serializable, UpdateableClassifier

public class NNge

extends Classifier

implements UpdateableClassifier, OptionHandler

NNge classifier. Nearest neighbor like algorithm using non-nested generalized exemplars. For more information, see

Brent Martin, (1995) "Instance-Based learning : Nearest Neighbor With Generalization", Master Thesis, University of Waikato, Hamilton, New Zealand

Sylvain Roy (2002) "Nearest Neighbor With Generalization", Unpublished, University of Canterbury, Christchurch, New Zealand

Valid options are:

-I num
Set the number of folder to use in the computing of the mutual information (default 5)

-G num
Set the number of attempts of generalisation (default 5)

Version:

$$

Author:

Brent Martin (bim20@cosc.canterbury.ac.nz), Sylvain Roy (sro33@student.canterbury.ac.nz)

See Also:

Serialized Form

Nested Class Summary

private class

NNge.Exemplar Implements Exemplar as used by NNge : parallel axis hyperrectangle.

Field Summary

private NNge.Exemplar	m_Exemplars The list of Exemplars
private NNge.Exemplar[]	m_ExemplarsByClass The lists of Exemplars by class
(package private) double[]	m_MaxArray The maximum values for numeric attributes.
private double[]	m_MI
private double[]	m_MI_MaxArray
private double[]	m_MI_MinArray
private int[][][]	m_MI_NumAttrClassInter MUTUAL INFORMATION'S DATAS
private int[][][]	m_MI_NumAttrClassValue
private int[][]	m_MI_NumAttrInter
private int[][]	m_MI_NumAttrValue
private int[]	m_MI_NumClass
private int	m_MI_NumInst
(package private) double[]	m_MinArray The minimum values for numeric attributes.
private double[]	m_MissingVector Values to use for missing value
private int	m_NumAttemptsOfGene The number of try for generalisation
private int	m_NumFoldersMI The number of folder for the Mutual Information
private Instances	m_Train An empty instances to keep the headers, the classIndex, etc...

Fields inherited from class weka.classifiers.Classifier

m_Debug

Constructor Summary

NNge()

Method Summary

private void	addExemplar(NNge.Exemplar ex) Adds an Exemplar in NNge's lists Ensure that the exemplar is not already in a list : the links would be broken...
private void	adjust(Instance newInst, NNge.Exemplar predictedExemplar) Adjust the NNge.
private double	attrWeight(int index) returns the weight of indexth attribute
void	buildClassifier(Instances data) Generates a classifier.
double	classifyInstance(Instance instance) Classifies a given instance.
private boolean	detectOverlapping(NNge.Exemplar ex) Returns true if ex overlaps any of the Exemplars in NNge's lists
private void	generalise(Instance newInst) Generalise an Exemplar (not necessarily predictedExemplar) to match instance.
int	getNumAttemptsOfGeneOption() Gets the number of attempts for generalisation.
int	getNumFoldersMIOption() Gets the number of folder for mutual information.
java.lang.String[]	getOptions() Gets the current option settings for the OptionHandler.
java.lang.String	globalInfo() Returns a string describing classifier
private void	initWeight(NNge.Exemplar ex) Init the weight of ex Watch out !
java.util.Enumeration	listOptions() Returns an enumeration of all the available options..
static void	main(java.lang.String[] argv) Main method for testing this class.
private NNge.Exemplar	nearestExemplar(Instance inst) Returns the nearest Exemplar
private NNge.Exemplar	nearestExemplar(Instance inst, double c) Returns the nearest Exemplar with class c
private boolean	notEqualFeatures(Instance inst1, Instance inst2) Returns true if the instance don't have the same feature values
java.lang.String	numAttemptsOfGeneOptionTipText() Returns the tip text for this property
java.lang.String	numFoldersMIOptionTipText() Returns the tip text for this property
private void	prune(NNge.Exemplar predictedExemplar, Instance newInst) Prunes an Exemplar that matches an Instance
private void	removeExemplar(NNge.Exemplar ex) Removes an Exemplar from NNge's lists Ensure that the Exemplar is actually in NNge's lists.
void	setNumAttemptsOfGeneOption(int newIntParameter) Sets the number of attempts for generalisation.
void	setNumFoldersMIOption(int newIntParameter) Sets the number of folder for mutual information.
void	setOptions(java.lang.String[] options) Sets the OptionHandler's options using the given list.
java.lang.String	toString() Returns a description of this classifier.
private void	update(Instance instance) Performs the update of the classifier
void	updateClassifier(Instance instance) Updates the classifier using the given instance.
private void	updateMI(Instance inst) Updates the data for computing the mutual information MUST be called AFTER adding inst in m_Train
private void	updateMinMax(Instance instance) Updates the minimum, maximum, sum, sumSquare values for all the attributes

Methods inherited from class weka.classifiers.Classifier

debugTipText, distributionForInstance, forName, getDebug, makeCopies, setDebug

Methods inherited from class java.lang.Object

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

Field Detail

m_Train

private Instances m_Train

An empty instances to keep the headers, the classIndex, etc...

m_Exemplars

private NNge.Exemplar m_Exemplars

The list of Exemplars

m_ExemplarsByClass

private NNge.Exemplar[] m_ExemplarsByClass

The lists of Exemplars by class

m_MinArray

double[] m_MinArray

The minimum values for numeric attributes.

m_MaxArray

double[] m_MaxArray

The maximum values for numeric attributes.

m_NumAttemptsOfGene

private int m_NumAttemptsOfGene

The number of try for generalisation

m_NumFoldersMI

private int m_NumFoldersMI

The number of folder for the Mutual Information

m_MissingVector

private double[] m_MissingVector

Values to use for missing value

m_MI_NumAttrClassInter

private int[][][] m_MI_NumAttrClassInter

MUTUAL INFORMATION'S DATAS

m_MI_NumAttrInter

private int[][] m_MI_NumAttrInter

m_MI_MaxArray

private double[] m_MI_MaxArray

m_MI_MinArray

private double[] m_MI_MinArray

m_MI_NumAttrClassValue

private int[][][] m_MI_NumAttrClassValue

m_MI_NumAttrValue

private int[][] m_MI_NumAttrValue

m_MI_NumClass

private int[] m_MI_NumClass

m_MI_NumInst

private int m_MI_NumInst

m_MI

private double[] m_MI

Constructor Detail

NNge

public NNge()

Method Detail

globalInfo

public java.lang.String globalInfo()

Returns a string describing classifier

Returns:

a description suitable for displaying in the explorer/experimenter gui

buildClassifier

public void buildClassifier(Instances data)
                     throws java.lang.Exception

Generates a classifier. Must initialize all fields of the classifier that are not being set via options (ie. multiple calls of buildClassifier must always lead to the same result). Must not change the dataset in any way.

Specified by:

buildClassifier in class Classifier

Parameters:

data - set of instances serving as training data

Throws:

java.lang.Exception - if the classifier has not been generated successfully

classifyInstance

public double classifyInstance(Instance instance)
                        throws java.lang.Exception

Classifies a given instance.

Overrides:

classifyInstance in class Classifier

Parameters:

instance - the instance to be classified

Returns:

index of the predicted class as a double

Throws:

java.lang.Exception - if instance could not be classified successfully

updateClassifier

public void updateClassifier(Instance instance)
                      throws java.lang.Exception

Updates the classifier using the given instance.

Specified by:

updateClassifier in interface UpdateableClassifier

Parameters:

instance - the instance to include

Throws:

java.lang.Exception - if instance could not be incorporated successfully

update

private void update(Instance instance)
             throws java.lang.Exception

Performs the update of the classifier

Parameters:

instance - the new instance

Throws:

java.lang.Exception - if the update fails

nearestExemplar

private NNge.Exemplar nearestExemplar(Instance inst)

Returns the nearest Exemplar

Parameters:

inst - an Instance

Returns:

the nearest Exemplar to inst, null if no exemplar are found.

nearestExemplar

private NNge.Exemplar nearestExemplar(Instance inst,
                                      double c)

Returns the nearest Exemplar with class c

Parameters:

inst - an Instance

c - the class of the Exemplar to return

Returns:

the nearest Exemplar to inst with class c, null if no exemplar with class c are found.

generalise

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

Generalise an Exemplar (not necessarily predictedExemplar) to match instance. predictedExemplar must be in NNge's lists

Parameters:

newInst - the new instance

Throws:

java.lang.Exception - in case of inconsitent situation

adjust

private void adjust(Instance newInst,
                    NNge.Exemplar predictedExemplar)
             throws java.lang.Exception

Adjust the NNge.

Parameters:

newInst - the instance to classify

predictedExemplar - the Exemplar that matches newInst

Throws:

java.lang.Exception - in case of inconsistent situation

prune

private void prune(NNge.Exemplar predictedExemplar,
                   Instance newInst)
            throws java.lang.Exception

Prunes an Exemplar that matches an Instance

Parameters:

predictedExemplar - an Exemplar

newInst - an Instance matched by predictedExemplar

Throws:

java.lang.Exception - in case of inconsistent situation. (shouldn't happen.)

notEqualFeatures

private boolean notEqualFeatures(Instance inst1,
                                 Instance inst2)

Returns true if the instance don't have the same feature values

Parameters:

inst1 - an instance

inst2 - an instance

Returns:

true if the instance don't have the same feature values

detectOverlapping

private boolean detectOverlapping(NNge.Exemplar ex)

Returns true if ex overlaps any of the Exemplars in NNge's lists

Parameters:

ex - an Exemplars

Returns:

true if ex overlaps any of the Exemplars in NNge's lists

updateMinMax

private void updateMinMax(Instance instance)

Updates the minimum, maximum, sum, sumSquare values for all the attributes

Parameters:

instance - the new instance

updateMI

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

Updates the data for computing the mutual information MUST be called AFTER adding inst in m_Train

Throws:

java.lang.Exception - is thrown if an inconsistent situation is met

initWeight

private void initWeight(NNge.Exemplar ex)

Init the weight of ex Watch out ! ex shouldn't be in NNge's lists when initialized

Parameters:

ex - the Exemplar to initialise

addExemplar

private void addExemplar(NNge.Exemplar ex)

Adds an Exemplar in NNge's lists Ensure that the exemplar is not already in a list : the links would be broken...

Parameters:

ex - a new Exemplar to add

removeExemplar

private void removeExemplar(NNge.Exemplar ex)

Removes an Exemplar from NNge's lists Ensure that the Exemplar is actually in NNge's lists. Likely to do something wrong if this condition is not respected. Due to the list implementation, the Exemplar can appear only once in the lists : once removed, the exemplar is not in the lists anymore.

Parameters:

ex - a new Exemplar to add

attrWeight

private double attrWeight(int index)

returns the weight of indexth attribute

Parameters:

index - attribute's index

Returns:

the weight of indexth attribute

toString

public java.lang.String toString()

Returns a description of this classifier.

Returns:

a description of this classifier as a string.

listOptions

public java.util.Enumeration listOptions()

Returns an enumeration of all the available options..

Specified by:

listOptions in interface OptionHandler

Overrides:

listOptions in class Classifier

Returns:

an enumeration of all available options.

setOptions

public void setOptions(java.lang.String[] options)
                throws java.lang.Exception

Sets the OptionHandler's options using the given list. All options will be set (or reset) during this call (i.e. incremental setting of options is not possible).

Specified by:

setOptions in interface OptionHandler

Overrides:

setOptions in class Classifier

Parameters:

options - the list of options as an array of strings

Throws:

java.lang.Exception - if an option is not supported

getOptions

public java.lang.String[] getOptions()

Gets the current option settings for the OptionHandler.

Specified by:

getOptions in interface OptionHandler

Overrides:

getOptions in class Classifier

Returns:

the list of current option settings as an array of strings

numAttemptsOfGeneOptionTipText

public java.lang.String numAttemptsOfGeneOptionTipText()

Returns the tip text for this property

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getNumAttemptsOfGeneOption

public int getNumAttemptsOfGeneOption()

Gets the number of attempts for generalisation.

Returns:

the value of the option G

setNumAttemptsOfGeneOption

public void setNumAttemptsOfGeneOption(int newIntParameter)

Sets the number of attempts for generalisation.

Parameters:

newIntParameter - the new value.

numFoldersMIOptionTipText

public java.lang.String numFoldersMIOptionTipText()

Returns the tip text for this property

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getNumFoldersMIOption

public int getNumFoldersMIOption()

Gets the number of folder for mutual information.

Returns:

the value of the option I

setNumFoldersMIOption

public void setNumFoldersMIOption(int newIntParameter)

Sets the number of folder for mutual information.

Parameters:

newIntParameter - the new value.

main

public static void main(java.lang.String[] argv)

Main method for testing this class.

Parameters:

argv - should contain command line arguments for evaluation (see Evaluation).