Package org.hortonmachine.gears.utils.clustering

Class GvmClusters<S extends GvmSpace,K>

java.lang.Object

org.hortonmachine.gears.utils.clustering.GvmClusters<S,K>

Type Parameters:

K - the key type

public class GvmClusters<S extends GvmSpace,K>
extends Object

Maintains a collection of clusters which are adjusted as new elements are added. The keyer assigned to this object should not be modified while there are clusters.

Author:

Tom Gibara

Constructor Summary

Constructors

Constructor	Description
GvmClusters(S space, int capacity)

Method Summary

Modifier and Type	Method	Description
void	add(double m, Object pt, K key)	Adds a point to be clustered.
void	clear()	Removes all clusters and clustered points but retains the keyer.
int	getCapacity()	The greatest number of clusters that may be recorded.
GvmKeyer<K>	getKeyer()	The keyer used to assign keys to clusters.
S	getSpace()	The space in which clustering will be performed
void	reduce(double maxVar, int minClusters)	Collapses the number of clusters subject to constraints on the maximum permitted variance, and the least number of clusters.
List<GvmResult<K>>	results()	Obtains the clusters for the points added.
void	setKeyer(GvmKeyer<K> keyer)	The keyer to be used.

Methods inherited from class java.lang.Object

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

Constructor Detail

GvmClusters

public GvmClusters(S space,
                   int capacity)

Method Detail

getKeyer

public GvmKeyer<K> getKeyer()

The keyer used to assign keys to clusters.

setKeyer

public void setKeyer(GvmKeyer<K> keyer)

The keyer to be used.

Parameters:

keyer - a keyer, not null

getCapacity

public int getCapacity()

The greatest number of clusters that may be recorded.

Returns:

the capacity

getSpace

public S getSpace()

The space in which clustering will be performed

Returns:

the space containing the points to be clustered

clear

public void clear()

Removes all clusters and clustered points but retains the keyer.

add

public void add(double m,
                Object pt,
                K key)

Adds a point to be clustered.

Parameters:

m - the mass at the point

xs - the coordinates of the point

key - a key assigned to the point, may be null

reduce

public void reduce(double maxVar,
                   int minClusters)

Collapses the number of clusters subject to constraints on the maximum permitted variance, and the least number of clusters. This method may be called at any time, including between calls to add().

Parameters:

maxVar - an upper bound on the global variance that may not be exceeded by merging clusters

minClusters - a lower bound on the the number of clusters that may not be exceeded by merging clusters

results

public List<GvmResult<K>> results()

Obtains the clusters for the points added. This method may be called at any time, including between calls to add().

Returns:

the result of clustering the points thus far added