Package ffx.numerics.clustering
@ParametersAreNonnullByDefault
package ffx.numerics.clustering
The clustering package provides hierarchical agglomerative clustering utilities,
including linkage strategies (single, complete, average, weighted), distance map
management, and supporting data structures and algorithms.
These classes are used to construct clustering trees from distance matrices and to produce flat or weighted clusterings as needed.
Original copyright and license: Copyright 2013 Lars Behnke
Apache License, Version 2.0
http://www.apache.org/licenses/LICENSE-2.0
- Since:
- 1.0
- Author:
- Lars Behnke, 2013, Michael J. Schnieders
-
ClassDescriptionLinkage strategy that uses the arithmetic mean of pairwise distances between cluster members (UPGMA/average linkage).Represents a node in a hierarchical clustering tree (dendrogram).Defines algorithms that produce hierarchical clusters from distance data.Immutable-like holder describing a pair of clusters and the linkage distance between them; used as entries within the DistanceMap during agglomeration.Linkage strategy that uses the maximum of pairwise distances between cluster members (complete-linkage / farthest-neighbor).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.Simple value object storing a distance and an optional weight used during linkage computations; comparable by distance and cloneable.Container for linkages with the minimal methods needed in the package Created by Alexandre Masselot on 7/18/14.Performs agglomerative steps to build a clustering hierarchy from an initial set of singleton clusters and a map of pairwise distances.Strategy interface for computing the inter-cluster distance used during agglomeration.Clustering algorithm that consumes a condensed (pdist-style) upper-triangular distance array to produce hierarchical agglomerative clusters.Linkage strategy that uses the minimum of pairwise distances between cluster members (single-linkage / nearest-neighbor).Linkage strategy that computes a weighted average of pairwise distances using the associated weights of cluster members.