keel.Algorithms.Semi_Supervised_Learning.utilities.KNN

Class KNN

java.lang.Object

keel.Algorithms.Semi_Supervised_Learning.utilities.KNN.KNN

public class KNN
extends java.lang.Object

Implements the KNN algorithm.

Author:

diegoj

Field Summary

Fields

Modifier and Type	Field and Description
protected static int	K Number of neighbor-prototypes to be searched in the KNN.

Constructor Summary

Constructors

Constructor and Description
KNN()

Method Summary

Modifier and Type	Method and Description
static Prototype	_1nn(Prototype current, PrototypeSet dataSet) Implements the 1NN algorithm
static int	classficationAccuracy(PrototypeSet training, PrototypeSet test) Informs of the classification accuracy
static int	classficationAccuracy(PrototypeSet training, PrototypeSet test, int k) Informs of the classification accuracy.
static int	classficationAccuracy1NN(PrototypeSet training, PrototypeSet test) Informs of the number of prototypes with correct class.
static Pair<java.lang.Integer,java.lang.Integer>	classficationAccuracyAndError1NN(PrototypeSet training, PrototypeSet test) Informs of the number of prototypes with correct class.
static int[]	classify(PrototypeSet training, PrototypeSet test, int k, double[][] probabilities) Informs of the classification done.
static int[]	classify2(PrototypeSet training, PrototypeSet test, int k) Informs of the classification done.
static int	getK() Returns the current value of K.
static Prototype	getNearest(Prototype current, PrototypeSet dataSet) Return the nearest prototype to another in a set.
static Prototype	getNearest(Prototype current, PrototypeSet dataSet, boolean isSameClass) Return the nearest prototype to another with the same of different class
static PrototypeSet	getNearestNeighbors(Prototype current, PrototypeSet dataSet, int numberOfNeighbors) Return some nearest prototypes to another with different class
static PrototypeSet	getNearestNeighborsWithDifferentClassAs(Prototype current, PrototypeSet dataSet, int numberOfNeighbors) Return some nearest prototypes to another with different class
static PrototypeSet	getNearestNeighborsWithSameClassAs(Prototype current, PrototypeSet dataSet) Return the nearest prototypes to another with the same class
static PrototypeSet	getNearestNeighborsWithSameClassAs(Prototype current, PrototypeSet dataSet, int numberOfNeighbors) Return some nearest prototypes to another with the same class
static Prototype	getNearestWithDifferentClassAs(Prototype current, PrototypeSet dataSet) Return the nearest prototype to another with the same class
static Prototype	getNearestWithSameClassAs(Prototype current, PrototypeSet dataSet) Return the nearest prototype to another with the same class.
static int	k() Returns the current value of K.
static PrototypeSet	knn(Prototype current, PrototypeSet dataSet) Implements the KNN algorithm.
static PrototypeSet	knn(Prototype current, PrototypeSet dataSet, int k) Implements the KNN algorithm
static void	setK(int n) Sets the number of prototypes to be used in the knn algorithm.

Methods inherited from class java.lang.Object

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

Field Detail

K

protected static int K

Number of neighbor-prototypes to be searched in the KNN.

Constructor Detail

KNN

public KNN()

Method Detail

setK

public static void setK(int n)

Sets the number of prototypes to be used in the knn algorithm. Must be integer greater than 0.

Parameters:

n - Number of prototypes

k

public static int k()

Returns the current value of K.

Returns:

Number of neighbors used with each prototype in the KNN.

getK

public static int getK()

Returns the current value of K.

Returns:

Number of neighbors used with each prototype in the KNN.

_1nn

public static Prototype _1nn(Prototype current,
                             PrototypeSet dataSet)

Implements the 1NN algorithm

Parameters:

current - Prototype which the algorithm will find its nearest-neighbor.

dataSet - Prototype set in which the algorithm will search.

Returns:

Nearest prototype to current in the prototype set dataset.

knn

public static PrototypeSet knn(Prototype current,
                               PrototypeSet dataSet,
                               int k)

Implements the KNN algorithm

Parameters:

current - Prototype which the algorithm will find its nearest-neighbors.

dataSet - Prototype set in which the algorithm will search.

k - The size neighborhood to be returned.

Returns:

Nearest prototypes to current in the prototype set dataset.

knn

public static PrototypeSet knn(Prototype current,
                               PrototypeSet dataSet)

Implements the KNN algorithm. Use static parameter of the class

Parameters:

current - Prototype which the algorithm will find its nearest-neighbors.

dataSet - Prototype set in which the algorithm will search.

Returns:

Nearest prototypes to current in the prototype set dataset.

classficationAccuracy1NN

public static int classficationAccuracy1NN(PrototypeSet training,
                                           PrototypeSet test)

Informs of the number of prototypes with correct class. Uses 1NN to perform nearest prototype.

Parameters:

training - Training data prototype set

test - Test data prototype set

Returns:

Number of prototypes well classificated

classficationAccuracyAndError1NN

public static Pair<java.lang.Integer,java.lang.Integer> classficationAccuracyAndError1NN(PrototypeSet training,
                                                                                         PrototypeSet test)

Informs of the number of prototypes with correct class. Uses 1NN to perform nearest prototype.

Parameters:

training - Training data prototype set

test - Test data prototype set

Returns:

Number of prototypes well classificated

classficationAccuracy

public static int classficationAccuracy(PrototypeSet training,
                                        PrototypeSet test,
                                        int k)

Informs of the classification accuracy. The number of prototypes must be at least 1.

Parameters:

training - Training data prototype set

test - Test data prototype set

k - Number of prototype-neighbors to be compared with each prototype of test data prototypes.

Returns:

Number of prototypes well classificated

classficationAccuracy

public static int classficationAccuracy(PrototypeSet training,
                                        PrototypeSet test)

Informs of the classification accuracy

Parameters:

training - Training data prototype set

test - Test data prototype set

Returns:

Number of prototypes well classificated

getNearest

public static Prototype getNearest(Prototype current,
                                   PrototypeSet dataSet,
                                   boolean isSameClass)

Return the nearest prototype to another with the same of different class

Parameters:

current - Prototype which the algorithm will find its nearest-neighbors.

dataSet - Prototype set in which the algorithm will search.

isSameClass - Must return a prototype with the same class as current?

Returns:

Nearest prototype to current with the same of different class (resp. isSameClass)

getNearest

public static Prototype getNearest(Prototype current,
                                   PrototypeSet dataSet)

Return the nearest prototype to another in a set.

Parameters:

current - Prototype which the algorithm will find its nearest-neighbors.

dataSet - Prototype set in which the algorithm will search.

Returns:

Nearest prototype to current in dataSet.

getNearestWithSameClassAs

public static Prototype getNearestWithSameClassAs(Prototype current,
                                                  PrototypeSet dataSet)

Return the nearest prototype to another with the same class.

Parameters:

current - Prototype which the algorithm will find its nearest-neighbor.

dataSet - Prototype set in which the algorithm will search.

Returns:

Nearest prototype to current with the same class.

getNearestNeighborsWithSameClassAs

public static PrototypeSet getNearestNeighborsWithSameClassAs(Prototype current,
                                                              PrototypeSet dataSet)

Return the nearest prototypes to another with the same class

Parameters:

current - Prototype which the algorithm will find its nearest-neighbor.

dataSet - Prototype set in which the algorithm will search.

Returns:

PrototypeSet containing all the prototypes of the same class in distance increasing order.

getNearestNeighborsWithSameClassAs

public static PrototypeSet getNearestNeighborsWithSameClassAs(Prototype current,
                                                              PrototypeSet dataSet,
                                                              int numberOfNeighbors)

Return some nearest prototypes to another with the same class

Parameters:

current - Prototype which the algorithm will find its nearest-neighbor.

dataSet - Prototype set in which the algorithm will search.

numberOfNeighbors - Number of neighbors to be returned.

Returns:

PrototypeSet containing all the prototypes of the same class in distance increasing order.

getNearestNeighborsWithDifferentClassAs

public static PrototypeSet getNearestNeighborsWithDifferentClassAs(Prototype current,
                                                                   PrototypeSet dataSet,
                                                                   int numberOfNeighbors)

Return some nearest prototypes to another with different class

Parameters:

current - Prototype which the algorithm will find its nearest-neighbor.

dataSet - Prototype set in which the algorithm will search.

numberOfNeighbors - Number of neighbors to be returned.

Returns:

PrototypeSet containing all the prototypes of the same class in distance increasing order.

getNearestWithDifferentClassAs

public static Prototype getNearestWithDifferentClassAs(Prototype current,
                                                       PrototypeSet dataSet)

Return the nearest prototype to another with the same class

Parameters:

current - Prototype which the algorithm will find its nearest-neighbor.

dataSet - Prototype set in which the algorithm will search.

Returns:

Nearest prototype to current with the same class.

getNearestNeighbors

public static PrototypeSet getNearestNeighbors(Prototype current,
                                               PrototypeSet dataSet,
                                               int numberOfNeighbors)

Return some nearest prototypes to another with different class

Parameters:

current - Prototype which the algorithm will find its nearest-neighbor.

dataSet - Prototype set in which the algorithm will search.

numberOfNeighbors - Number of neighbors to be returned.

Returns:

PrototypeSet containing all the prototypes of the same class in distance increasing order.

classify

public static int[] classify(PrototypeSet training,
                             PrototypeSet test,
                             int k,
                             double[][] probabilities)

Informs of the classification done. The number of prototypes must be at least 1.

Parameters:

training - Training data prototype set

test - Test data prototype set

k - Number of prototype-neighbors to be compared with each prototype of test data prototypes.

probabilities - classes probabilities for the each test instance.

Returns:

predicted classes of the test instances.

classify2

public static int[] classify2(PrototypeSet training,
                              PrototypeSet test,
                              int k)

Informs of the classification done. The number of prototypes must be at least 1.

Parameters:

training - Training data prototype set

test - Test data prototype set

k - Number of prototype-neighbors to be compared with each prototype of test data prototypes.

Returns:

Number of prototypes well classificated