Crates.io | fuzzy_dbscan |
lib.rs | fuzzy_dbscan |
version | 0.3.0 |
source | src |
created_at | 2018-07-19 20:52:03.491259 |
updated_at | 2018-12-06 23:16:20.137141 |
description | An implementation of the FuzzyDBSCAN algorithm |
homepage | https://github.com/schulzch/fuzzy_dbscan |
repository | https://github.com/schulzch/fuzzy_dbscan |
max_upload_size | |
id | 75126 |
size | 254,918 |
An implementation of the FuzzyDBSCAN algorithm [1].
FuzzyDBSCAN is an agglomerative fuzzy clustering algorithm that groups a set of points in such a way that one point can belong to more than one group. The assignment of a point is expressed as a category (core, border, noise) and a soft label (between 0.0 and 1.0). Only points from the border category can be assigned ambiguously.
An example of two slightly connected clusters (red and blue) where the transition is assigned to both clusters, i.e., fuzzy (red + blue = purple). Core points have an enclosing circle, whereas border points do not. The opacity indicates the degree of membership, i.e., soft label.
See documentation for an example.
Compile the crate to WASM using wasm-pack build --target=nodejs
(or --target=browser
), then run it using:
var FuzzyDBSCAN = require('fuzzy_dbscan.js'); // only for Node.js
var fuzzyDBSCAN = new FuzzyDBSCAN.FuzzyDBSCAN();
fuzzyDBSCAN.eps_min = 10.0;
fuzzyDBSCAN.eps_max = 20.0;
fuzzyDBSCAN.pts_min = 1;
fuzzyDBSCAN.pts_max = 2;
console.log(fuzzyDBSCAN.cluster([{x: 0, y: 0}, {x: 100, y: 100}, {x: 105, y: 105}, {x: 115, y: 115}]));
[1] Dino Ienco, and Gloria Bordogna. "Fuzzy extensions of the DBScan clustering algorithm." Soft Computing (2016).
This project is maintained under the Semantic Versioning guidelines.
Licensed under the Apache 2.0 License. Copyright © 2018 Christoph Schulz.