2.3.3 cluster

The cluster program reads an unclustered dataset from standard input, such as one created by the generate program. It then runs the specified clustering algorithm on this dataset, and outputs the resultant clustered dataset to standard output. The representatives of clusters are also output, and the annotations of the dataset points are preserved.

The algorithms themselves will be detailed later.

Usage:

cluster [OPTIONS]

-help

Outputs a brief usage message.

-debug

Output debugging information.

-algorithm algorithmname [k]

Specifies the clustering algorithm to use, and optionally a suggestion for the number of clusters () for those algorithms which need or benefit from it. algorithm_name defaults to ``kmeans'', defaults to 4.

Available algorithms:

kmeans

The standard -MEANS algorithm, Section 3.3.

tb

The Teitz-Bart -MEDOIDS heuristic, Section 3.4.

ech

The ECH variant of the TB -MEDOIDS heuristic, Section 3.5.

random

Random clustering (equal sized clusters), Section 3.2.

cheat

Cheating clustering, using the annotation, Section 3.1.3.

ascending

Puts each point into its own cluster, Section 3.1.2.

single

Puts all points into a single cluster, Section 3.1.1.