keel.Algorithms.Preprocess.Missing_Values.kmeansImpute

Class gCenter

java.lang.Object

keel.Algorithms.Preprocess.Missing_Values.kmeansImpute.gCenter

public class gCenter
extends java.lang.Object

This class represents a group of centers (centroids) of a set of clusters

Constructor Summary

Constructors

Constructor and Description
gCenter() Creates a new instance of gCenter
gCenter(int k, int ndatos, int nvariables) Creates a new instance of gCenter with provided number of centers, number of instances of the data set and number of attributes

Method Summary

Modifier and Type	Method and Description
void	copyCenter(Instance i, int c) this function initializes a center with the values of a given instance.
double	distance(Instance i, int k) Computes the distance between a instances (without previous normalization) and one clusters (i.e. its centroid).
int	getClusterOf(int orderOf_i) Returns the cluster to which the given instance belongs to
int	nearestCenter(Instance inst) Computes the nearest cluster to the given instance
void	recalculateCenters(InstanceSet IS) Recalculates all the centroids using a given InstanceSet, in order to reduce the total sum of distances for each object to the centroid of the cluster, which the object belongs to
void	setClusterOf(Instance i, int orderOf_i) Updates the cluster membership of the instance to the nearest cluster
java.lang.String	valueAt(int cluster, int position) Get the value of an attribute of the indicated centroid

Methods inherited from class java.lang.Object

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

Constructor Detail

gCenter

public gCenter()

Creates a new instance of gCenter

gCenter

public gCenter(int k,
               int ndatos,
               int nvariables)

Creates a new instance of gCenter with provided number of centers, number of instances of the data set and number of attributes

Parameters:

k - Fixed number of centroids

ndatos - number of instances in the data set related to this object

nvariables - number of attributes

Method Detail

distance

public double distance(Instance i,
                       int k)

Computes the distance between a instances (without previous normalization) and one clusters (i.e. its centroid).

Parameters:

i - The reference instance

k - The cluster number

Returns:

The Euclidean distance between i and k

nearestCenter

public int nearestCenter(Instance inst)

Computes the nearest cluster to the given instance

Parameters:

inst - The instance we are interested to compare

Returns:

The index of the nearest cluster

copyCenter

public void copyCenter(Instance i,
                       int c)

this function initializes a center with the values of a given instance.

Parameters:

i - the initialization instance

c - the index of the cluster to be initialized

recalculateCenters

public void recalculateCenters(InstanceSet IS)

Recalculates all the centroids using a given InstanceSet, in order to reduce the total sum of distances for each object to the centroid of the cluster, which the object belongs to

Parameters:

IS - The reference InstanceSet

setClusterOf

public void setClusterOf(Instance i,
                         int orderOf_i)

Updates the cluster membership of the instance to the nearest cluster

Parameters:

i - The considered instance

orderOf_i - The index of the instance i

getClusterOf

public int getClusterOf(int orderOf_i)

Returns the cluster to which the given instance belongs to

Parameters:

orderOf_i - The index of the instance

Returns:

The index of the cluster to this isntance belongs to.

valueAt

public java.lang.String valueAt(int cluster,
                                int position)

Get the value of an attribute of the indicated centroid

Parameters:

cluster - The index of the cluster (centroid)

position - The attribute (dimension) to be obtained

Returns:

the current value of the dimension of the given cluster