Package ffx.numerics.clustering

Class DefaultClusteringAlgorithm

java.lang.Object

ffx.numerics.clustering.DefaultClusteringAlgorithm

All Implemented Interfaces:

ClusteringAlgorithm

public class DefaultClusteringAlgorithm
extends Object
implements ClusteringAlgorithm

Clustering algorithm that operates on a full N x N distance matrix to produce hierarchical agglomerative clusters (dendrogram), with optional support for per-element weights and flat clustering by threshold.

Since:

1.0

Author:

Lars Behnke, 2013, Michael J. Schnieders

Constructor Summary

Constructors

Constructor	Description
DefaultClusteringAlgorithm()

Method Summary

Modifier and Type	Method	Description
Cluster	performClustering(double[][] distances, String[] clusterNames, LinkageStrategy linkageStrategy)	Performs hierarchical agglomerative clustering using a full N x N distance matrix.
List<Cluster>	performFlatClustering(double[][] distances, String[] clusterNames, LinkageStrategy linkageStrategy, Double threshold)	Produces a flat clustering by agglomerating until the next merge would exceed the threshold.
Cluster	performWeightedClustering(double[][] distances, String[] clusterNames, double[] weights, LinkageStrategy linkageStrategy)	Performs hierarchical clustering when each element has an associated weight.

Methods inherited from class java.lang.Object

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

Constructor Details

DefaultClusteringAlgorithm

public DefaultClusteringAlgorithm()

Method Details

performClustering

public Cluster performClustering(double[][] distances,
 String[] clusterNames,
 LinkageStrategy linkageStrategy)

Performs hierarchical agglomerative clustering using a full N x N distance matrix.

Specified by:

performClustering in interface ClusteringAlgorithm

Parameters:

distances - an N x N symmetric matrix of pairwise distances

clusterNames - names corresponding to rows/columns of the distance matrix

linkageStrategy - linkage criterion used to compute inter-cluster distances

Returns:

root Cluster of the resulting hierarchy

performFlatClustering

public List<Cluster> performFlatClustering(double[][] distances,
 String[] clusterNames,
 LinkageStrategy linkageStrategy,
 Double threshold)

Produces a flat clustering by agglomerating until the next merge would exceed the threshold.

Specified by:

performFlatClustering in interface ClusteringAlgorithm

Parameters:

distances - an N x N symmetric matrix of pairwise distances

clusterNames - names corresponding to the distance matrix

linkageStrategy - linkage criterion used during agglomeration

threshold - maximum allowed linkage distance for merging

Returns:

list of clusters at the chosen cut

performWeightedClustering

public Cluster performWeightedClustering(double[][] distances,
 String[] clusterNames,
 double[] weights,
 LinkageStrategy linkageStrategy)

Performs hierarchical clustering when each element has an associated weight.

Specified by:

performWeightedClustering in interface ClusteringAlgorithm

Parameters:

distances - an N x N symmetric matrix of distances

clusterNames - names for the N input elements

weights - weights for the N input elements

linkageStrategy - linkage criterion to use

Returns:

root Cluster of the resulting hierarchy