lerouxrgd.clj-hclust

https://github.com/lerouxrgd/clj-hclust.git

git clone 'https://github.com/lerouxrgd/clj-hclust.git'

(ql:quickload :lerouxrgd.clj-hclust)
5

clj-hclust

A hierarchical clustering and visualisation library for clojure based on core.matrix.

Clojars Project

Usage

All clustering results are stored in vectors with the form [left right distance]. A leaf is simply [id-self id-self 0] where id-self is its index in the matrix.

Here is a simple example:

(require '[clj-hclust.core :as hc]
         '[clojure.core.matrix :as m])

(def M (m/matrix
        [[0.00 0.50 2.24 3.35 3.00]
         [0.50 0.00 2.50 3.61 3.04]
         [2.24 2.50 0.00 1.12 1.41]
         [3.35 3.61 1.12 0.00 1.50]
         [3.00 3.04 1.41 1.50 0.00]]))
    
(hc/hclust-lw M :single-link)

;;=> [[[0 0 0] [1 1 0] 0.5] [[[2 2 0] [3 3 0] 1.12] [4 4 0] 1.41] 2.24]

Under the hood core.matrix is used with :vectorz as default implementations. You can use another implementation by adding corresponding dependency to your project file and selecting implementation using set-current-implementation. Note that the selected implementation must support emap! from the core.matrix API.

Algorithms

Lance-Williams

The Lance–Williams algorithms are an infinite family of agglomerative hierarchical clustering algorithms which are represented by a recursive formula for updating cluster distances at each step (each time a pair of clusters is merged).

The multimethod lw-updater allows for custom implementations of such algorithms. Available implementations are:

Visualization

SVG format

A simple batik based JFrame visualisation is available (note that it's possible to get just the SVG string):

(require '[clj-hclust.batik :as b])

(-> M
    (hc/hclust-lw :single-link)
    (hc/hclust->svg)
    (b/svg-jframe 600 100))

SVG visualisation parameters can be customized:

(-> M
    (hc/hclust-lw :single-link)
    (hc/hclust->svg {:names ["Rochefort" "Milady" "Athos" "Portos" "Aramis"]
                     :circle-style {:r 10}
                     :text-style {:dx 16 :dy 6 :font "14px sans-serif"}})
    (b/svg-jframe 700 150))

Newick format

A newick format visualisation is also available:

(-> M
    (hc/hclust-lw :single-link)
    (hc/hclust->newick))

;;=> "(((0:0,1:0):0.5,((2:0,3:0):1.12,4:0):1.41):2.24);"

License

Copyright © Romain Leroux

This project is licensed under the Eclipse Public License 1.0.