Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/lerouxrgd/clj-hclust
A hierarchical clustering and visualisation library for clojure based on core.matrix
https://github.com/lerouxrgd/clj-hclust
clojure clustering
Last synced: about 2 months ago
JSON representation
A hierarchical clustering and visualisation library for clojure based on core.matrix
- Host: GitHub
- URL: https://github.com/lerouxrgd/clj-hclust
- Owner: lerouxrgd
- License: epl-1.0
- Created: 2015-07-22T12:38:48.000Z (over 9 years ago)
- Default Branch: master
- Last Pushed: 2019-10-10T00:16:42.000Z (about 5 years ago)
- Last Synced: 2024-07-12T13:29:53.762Z (6 months ago)
- Topics: clojure, clustering
- Language: Clojure
- Homepage:
- Size: 30.3 KB
- Stars: 6
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# clj-hclust
A [hierarchical clustering][hc] and visualisation library for clojure based on core.matrix.
[](https://clojars.org/clj-hclust)
## 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:
```clj
(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][lance-williams] are an infinite family of agglomerative [hierarchical clustering][hc] 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:
* `:single-link` uses min when merging
* `:complete-link` uses max when merging
* `:ward` minimizes the total within-cluster variance when merging
## Visualization
### SVG format
A simple [batik][apache-batik] based `JFrame` visualisation is available (note that it's possible to get just the SVG string):
```clj
(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:
```clj
(-> 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][newick] visualisation is also available:
```clj
(-> 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][license].
[hc]: https://en.wikipedia.org/wiki/Hierarchical_clustering
[lance-williams]: https://en.wikipedia.org/wiki/Ward%27s_method#Lance.E2.80.93Williams_algorithms
[apache-batik]: https://xmlgraphics.apache.org/batik/
[newick]: https://en.wikipedia.org/wiki/Newick_format
[license]: http://www.eclipse.org/legal/epl-v10.html