Package elki.clustering.kmeans.initialization.betula
Initialization methods for BIRCHbased kmeans and EM clustering.

Interface Summary Interface Description CFInitWeight Initialization weight function for kmeans initialization with BETULA. 
Class Summary Class Description AbstractCFKMeansInitialization Abstract base class for CF kmeans initializations.AbstractCFKMeansInitialization.Par Parameterization class.CFKPlusPlusLeaves KMeans++like initialization for BETULA kmeans, treating the leaf clustering features as a flat list, and called "leaves" in the publication.CFKPlusPlusLeaves.Par Parameterization class.CFKPlusPlusTree Initialize Kmeans by following tree paths weighted by their variance contribution.CFKPlusPlusTree.Par Parameterization class.CFKPlusPlusTrunk Trunk strategy for initializing kmeans with BETULA: only the nodes up to a particular level are considered for kmeans++ style initialization.CFKPlusPlusTrunk.Par Parameterization class.CFRandomlyChosen Initialize Kmeans by randomly choosing k existing elements as initial cluster centers for Clustering Features.CFRandomlyChosen.Par Parameterization class.CFWeightedRandomlyChosen Initialize Kmeans by randomly choosing k existing elements as initial cluster centers for Clustering Features.CFWeightedRandomlyChosen.Par Parameterization class.InterclusterWeight Initialization via n2 * D2²(cf1, cf2), which supposedly is closes to the idea of kmeans++ initialization.SquaredEuclideanWeight Use the squared Euclidean distance only for distance measurement.VarianceWeight Variancebased weighting scheme for kmeans clustering with BETULA.