weka.classifiers.trees.j48
Class Distribution

java.lang.Object
  weka.classifiers.trees.j48.Distribution

All Implemented Interfaces:

java.lang.Cloneable, java.io.Serializable

public class Distribution

extends java.lang.Object

implements java.lang.Cloneable, java.io.Serializable

Class for handling a distribution of class values.

Version:

$Revision: 1.7 $

Author:

Eibe Frank (eibe@cs.waikato.ac.nz)

See Also:

Serialized Form

Field Summary

private double[]	m_perBag Weight of instances per bag.
private double[]	m_perClass Weight of instances per class.
private double[][]	m_perClassPerBag Weight of instances per class per bag.
private double	totaL Total weight of instances.

Constructor Summary

Distribution(Distribution toMerge)
Creates distribution with only one bag by merging all bags of given distribution.

Distribution(Distribution toMerge, int index)
Creates distribution with two bags by merging all bags apart of the indicated one.

Distribution(double[][] table)
Creates and initializes a new distribution using the given array.

Distribution(Instances source)
Creates a distribution with only one bag according to instances in source.

Distribution(Instances source, ClassifierSplitModel modelToUse)
Creates a distribution according to given instances and split model.

Distribution(int numBags, int numClasses)
Creates and initializes a new distribution.

Method Summary

int	actualNumBags() Returns number of non-empty bags of distribution.
int	actualNumClasses() Returns number of classes actually occuring in distribution.
int	actualNumClasses(int bagIndex) Returns number of classes actually occuring in given bag.
void	add(int bagIndex, double[] counts) Adds counts to given bag.
void	add(int bagIndex, Instance instance) Adds given instance to given bag.
void	addInstWithUnknown(Instances source, int attIndex) Adds all instances with unknown values for given attribute, weighted according to frequency of instances in each bag.
void	addRange(int bagIndex, Instances source, int startIndex, int lastPlusOne) Adds all instances in given range to given bag.
void	addWeights(Instance instance, double[] weights) Adds given instance to all bags weighting it according to given weights.
boolean	check(double minNoObj) Checks if at least two bags contain a minimum number of instances.
java.lang.Object	clone() Clones distribution (Deep copy of distribution).
void	del(int bagIndex, Instance instance) Deletes given instance from given bag.
void	delRange(int bagIndex, Instances source, int startIndex, int lastPlusOne) Deletes all instances in given range from given bag.
java.lang.String	dumpDistribution() Prints distribution.
void	initialize() Sets all counts to zero.
double	laplaceProb(int classIndex) Returns relative frequency of class over all bags with Laplace correction.
double	laplaceProb(int classIndex, int intIndex) Returns relative frequency of class for given bag.
double[][]	matrix() Returns matrix with distribution of class values.
int	maxBag() Returns index of bag containing maximum number of instances.
int	maxClass() Returns class with highest frequency over all bags.
int	maxClass(int index) Returns class with highest frequency for given bag.
int	numBags() Returns number of bags.
int	numClasses() Returns number of classes.
double	numCorrect() Returns perClass(maxClass()).
double	numCorrect(int index) Returns perClassPerBag(index,maxClass(index)).
double	numIncorrect() Returns total-numCorrect().
double	numIncorrect(int index) Returns perBag(index)-numCorrect(index).
double	perBag(int bagIndex) Returns number of (possibly fractional) instances in given bag.
double	perClass(int classIndex) Returns number of (possibly fractional) instances of given class.
double	perClassPerBag(int bagIndex, int classIndex) Returns number of (possibly fractional) instances of given class in given bag.
double	prob(int classIndex) Returns relative frequency of class over all bags.
double	prob(int classIndex, int intIndex) Returns relative frequency of class for given bag.
void	shift(int from, int to, Instance instance) Shifts given instance from one bag to another one.
void	shiftRange(int from, int to, Instances source, int startIndex, int lastPlusOne) Shifts all instances in given range from one bag to another one.
void	sub(int bagIndex, Instance instance) Subtracts given instance from given bag.
Distribution	subtract(Distribution toSubstract) Subtracts the given distribution from this one.
double	total() Returns total number of (possibly fractional) instances.

Methods inherited from class java.lang.Object

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

Field Detail

m_perClassPerBag

private double[][] m_perClassPerBag

Weight of instances per class per bag.

m_perBag

private double[] m_perBag

Weight of instances per bag.

m_perClass

private double[] m_perClass

Weight of instances per class.

totaL

private double totaL

Total weight of instances.

Constructor Detail

Distribution

public Distribution(int numBags,
                    int numClasses)

Creates and initializes a new distribution.

Distribution

public Distribution(double[][] table)

Creates and initializes a new distribution using the given array. WARNING: it just copies a reference to this array.

Distribution

public Distribution(Instances source)
             throws java.lang.Exception

Creates a distribution with only one bag according to instances in source.

Throws:

java.lang.Exception - if something goes wrong

Distribution

public Distribution(Instances source,
                    ClassifierSplitModel modelToUse)
             throws java.lang.Exception

Creates a distribution according to given instances and split model.

Throws:

java.lang.Exception - if something goes wrong

Distribution

public Distribution(Distribution toMerge)

Creates distribution with only one bag by merging all bags of given distribution.

Distribution

public Distribution(Distribution toMerge,
                    int index)

Creates distribution with two bags by merging all bags apart of the indicated one.

Method Detail

actualNumBags

public final int actualNumBags()

Returns number of non-empty bags of distribution.

actualNumClasses

public final int actualNumClasses()

Returns number of classes actually occuring in distribution.

actualNumClasses

public final int actualNumClasses(int bagIndex)

Returns number of classes actually occuring in given bag.

add

public final void add(int bagIndex,
                      Instance instance)
               throws java.lang.Exception

Adds given instance to given bag.

Throws:

java.lang.Exception - if something goes wrong

sub

public final void sub(int bagIndex,
                      Instance instance)
               throws java.lang.Exception

Subtracts given instance from given bag.

Throws:

java.lang.Exception - if something goes wrong

add

public final void add(int bagIndex,
                      double[] counts)

Adds counts to given bag.

addInstWithUnknown

public final void addInstWithUnknown(Instances source,
                                     int attIndex)
                              throws java.lang.Exception

Adds all instances with unknown values for given attribute, weighted according to frequency of instances in each bag.

Throws:

java.lang.Exception - if something goes wrong

addRange

public final void addRange(int bagIndex,
                           Instances source,
                           int startIndex,
                           int lastPlusOne)
                    throws java.lang.Exception

Adds all instances in given range to given bag.

Throws:

java.lang.Exception - if something goes wrong

addWeights

public final void addWeights(Instance instance,
                             double[] weights)
                      throws java.lang.Exception

Adds given instance to all bags weighting it according to given weights.

Throws:

java.lang.Exception - if something goes wrong

check

public final boolean check(double minNoObj)

Checks if at least two bags contain a minimum number of instances.

clone

public final java.lang.Object clone()

Clones distribution (Deep copy of distribution).

del

public final void del(int bagIndex,
                      Instance instance)
               throws java.lang.Exception

Deletes given instance from given bag.

Throws:

java.lang.Exception - if something goes wrong

delRange

public final void delRange(int bagIndex,
                           Instances source,
                           int startIndex,
                           int lastPlusOne)
                    throws java.lang.Exception

Deletes all instances in given range from given bag.

Throws:

java.lang.Exception - if something goes wrong

dumpDistribution

public final java.lang.String dumpDistribution()

Prints distribution.

initialize

public final void initialize()

Sets all counts to zero.

matrix

public final double[][] matrix()

Returns matrix with distribution of class values.

maxBag

public final int maxBag()

Returns index of bag containing maximum number of instances.

maxClass

public final int maxClass()

Returns class with highest frequency over all bags.

maxClass

public final int maxClass(int index)

Returns class with highest frequency for given bag.

numBags

public final int numBags()

Returns number of bags.

numClasses

public final int numClasses()

Returns number of classes.

numCorrect

public final double numCorrect()

Returns perClass(maxClass()).

numCorrect

public final double numCorrect(int index)

Returns perClassPerBag(index,maxClass(index)).

numIncorrect

public final double numIncorrect()

Returns total-numCorrect().

numIncorrect

public final double numIncorrect(int index)

Returns perBag(index)-numCorrect(index).

perClassPerBag

public final double perClassPerBag(int bagIndex,
                                   int classIndex)

Returns number of (possibly fractional) instances of given class in given bag.

perBag

public final double perBag(int bagIndex)

Returns number of (possibly fractional) instances in given bag.

perClass

public final double perClass(int classIndex)

Returns number of (possibly fractional) instances of given class.

laplaceProb

public final double laplaceProb(int classIndex)

Returns relative frequency of class over all bags with Laplace correction.

laplaceProb

public final double laplaceProb(int classIndex,
                                int intIndex)

Returns relative frequency of class for given bag.

prob

public final double prob(int classIndex)

Returns relative frequency of class over all bags.

prob

public final double prob(int classIndex,
                         int intIndex)

Returns relative frequency of class for given bag.

subtract

public final Distribution subtract(Distribution toSubstract)

Subtracts the given distribution from this one. The results has only one bag.

total

public final double total()

Returns total number of (possibly fractional) instances.

shift

public final void shift(int from,
                        int to,
                        Instance instance)
                 throws java.lang.Exception

Shifts given instance from one bag to another one.

Throws:

java.lang.Exception - if something goes wrong

shiftRange

public final void shiftRange(int from,
                             int to,
                             Instances source,
                             int startIndex,
                             int lastPlusOne)
                      throws java.lang.Exception

Shifts all instances in given range from one bag to another one.

Throws:

java.lang.Exception - if something goes wrong