Package elki.clustering.kmedoids

Class PAM.Instance

java.lang.Object

elki.clustering.kmedoids.PAM.Instance

Direct Known Subclasses:

EagerPAM.Instance, FastPAM1.Instance, ReynoldsPAM.Instance

Enclosing class:

PAM<O>

protected static class PAM.Instance
extends java.lang.Object

Instance for a single dataset.

Note: we experimented with not caching the distance to nearest and second nearest, but only the assignments. The matrix lookup was more expensive, so this is probably worth the 2*n doubles in storage.

Author:

Erich Schubert

Field Summary

Fields

Modifier and Type	Field	Description
(package private) WritableIntegerDataStore	assignment	Cluster mapping.
(package private) DistanceQuery<?>	distQ	Distance function to use.
(package private) DBIDs	ids	Ids to process.
(package private) WritableDoubleDataStore	nearest	Distance to the nearest medoid of each point.
(package private) WritableDoubleDataStore	second	Distance to the second nearest medoid.

Constructor Summary

Constructors

Constructor	Description
Instance(DistanceQuery<?> distQ, DBIDs ids, WritableIntegerDataStore assignment)	Constructor.

Method Summary

Modifier and Type	Method	Description
protected double	assignToNearestCluster(ArrayDBIDs means)	Assign each object to the nearest cluster, return the cost.
protected double	computeReassignmentCost(DBIDRef h, int mnum)	Compute the reassignment cost of one swap.
protected double	run(ArrayModifiableDBIDs medoids, int maxiter)	Run the PAM optimization phase.

Methods inherited from class java.lang.Object

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

Field Detail

ids

DBIDs ids

Ids to process.

distQ

DistanceQuery<?> distQ

Distance function to use.

nearest

WritableDoubleDataStore nearest

Distance to the nearest medoid of each point.

second

WritableDoubleDataStore second

Distance to the second nearest medoid.

assignment

WritableIntegerDataStore assignment

Cluster mapping.

Constructor Detail

Instance

public Instance(DistanceQuery<?> distQ,
                DBIDs ids,
                WritableIntegerDataStore assignment)

Constructor.

Parameters:

distQ - Distance query

ids - IDs to process

assignment - Cluster assignment

Method Detail

run

protected double run(ArrayModifiableDBIDs medoids,
                     int maxiter)

Run the PAM optimization phase.

Parameters:

medoids - Initial medoids list

maxiter - Maximum number of iterations

Returns:

final cost

computeReassignmentCost

protected double computeReassignmentCost(DBIDRef h,
                                         int mnum)

Compute the reassignment cost of one swap.

Parameters:

h - Current object to swap with the medoid

mnum - Medoid number to be replaced

Returns:

cost

assignToNearestCluster

protected double assignToNearestCluster(ArrayDBIDs means)

Assign each object to the nearest cluster, return the cost.

Parameters:

means - Object centroids

Returns:

Assignment cost