Expand description

DBSCAN Clustering

Note: This module is likely to change dramatically in the future and should be treated as experimental.

Provides an implementaton of DBSCAN clustering. The model also implements a predict function which uses nearest neighbours to classify the points. To utilize this function you must use self.set_predictive(true) before training the model.

The algorithm works by specifying eps and min_points parameters. The eps parameter controls how close together points must be to be placed in the same cluster. The min_points parameter controls how many points must be within distance eps of eachother to be considered a cluster.

If a point is not within distance eps of a cluster it will be classified as noise. This means that it will be set to None in the clusters Vector.

Examples

use rusty_machine::learning::dbscan::DBSCAN;
use rusty_machine::learning::UnSupModel;
use rusty_machine::linalg::Matrix;

let inputs = Matrix::new(6, 2, vec![1.0, 2.0,
                                    1.1, 2.2,
                                    0.9, 1.9,
                                    1.0, 2.1,
                                    -2.0, 3.0,
                                    -2.2, 3.1]);

let mut model = DBSCAN::new(0.5, 2);
model.train(&inputs).unwrap();

let clustering = model.clusters().unwrap();

Structs

DBSCAN Model