A few clustering methods including K-Means++ and UPGMA.

K-Means operates on an N-dimensional point type. Three initializers are provided, "++", random, and first points in list. A K-means++ wrapper is provided as a convenience.

A soft K-Means variant uses expectation maximization. This also operates on the same N-dimensional point type.

The hierarchical methods here take a distance matrix as input. The method Ultrametric can perform either UPGMA or single-linkage clustering and produce a rooted ultrametric tree. Methods AdditiveTree and NeighborJoin produce unrooted binary trees.

CAST stands for Cluster Affinity Search Technique. It clusters points where linking points with a similarity threshold would approximate a clique graph.

A few methods are exported for the Point type, including a Pearson correlation coefficient function useful for constructing similarity matrices. Also some data validation methods, a random tree generator and a random distance matrix generator.