edu.ucla.sspace.clustering.seeding

Interface KMeansSeed

All Known Implementing Classes:

GeneralizedOrssSeed, KMeansPlusPlusSeed, OrssSeed, RandomSeed

public interface KMeansSeed

An interface for KMeans seeding algorithms. Implementations must compose initial centroid seeds from a data set by either choosing an already existing data point or composing a linear combination of existing data points.

Implementations must be state free and threadsafe.

Author:

Keith Stevens

Method Summary

Methods

Modifier and Type	Method and Description
DoubleVector[]	chooseSeeds(int numCentroids, Matrix dataPoints) Returns an array of length numCentroids that contains centroids composed of either vectors from dataPoints or a linear combination of vectors from dataPoints.

Method Detail

chooseSeeds

DoubleVector[] chooseSeeds(int numCentroids,
                         Matrix dataPoints)

Returns an array of length numCentroids that contains centroids composed of either vectors from dataPoints or a linear combination of vectors from dataPoints.