moa.clusterers.outliers.utils.mtree

Class PartitionFunctions.BalancedPartition<DATA>

java.lang.Object

moa.clusterers.outliers.utils.mtree.PartitionFunctions.BalancedPartition<DATA>

Type Parameters:

DATA - The type of the data objects.

All Implemented Interfaces:

PartitionFunction<DATA>

Enclosing class:

PartitionFunctions

public static class PartitionFunctions.BalancedPartition<DATA>
extends Object
implements PartitionFunction<DATA>

A partition function that tries to distribute the data objects equally between the promoted data objects, associating to each promoted data objects the nearest data objects.

Constructor Summary

Constructors

Constructor and Description
PartitionFunctions.BalancedPartition()

Method Summary

Methods

Modifier and Type	Method and Description
Pair<Set<DATA>>	process(Pair<DATA> promoted, Set<DATA> dataSet, DistanceFunction<? super DATA> distanceFunction) Processes the balanced partition.

Methods inherited from class java.lang.Object

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

Constructor Detail

PartitionFunctions.BalancedPartition

public PartitionFunctions.BalancedPartition()

Method Detail

process

public Pair<Set<DATA>> process(Pair<DATA> promoted,
                      Set<DATA> dataSet,
                      DistanceFunction<? super DATA> distanceFunction)

Processes the balanced partition.

The algorithm is roughly equivalent to this:

     While dataSet is not Empty:
         X := The object in dataSet which is nearest to promoted.first
         Remove X from dataSet
         Add X to result.first
         
         Y := The object in dataSet which is nearest to promoted.second
         Remove Y from dataSet
         Add Y to result.second
         
     Return result
 

Specified by:

process in interface PartitionFunction<DATA>

Parameters:

promoted - The pair of data objects that will guide the partition process.

dataSet - The original set of data objects to be partitioned.

distanceFunction - A distance function to be used on the partitioning.

Returns:

A pair of partition sub-sets. Each sub-set must correspond to one of the promoted data objects.

See Also:

mtree.PartitionFunction#process(mtree.utils.Pair, java.util.Set, mtree.DistanceFunction)