/dbscan-python

Python implementation of 'DBSCAN' Algorithm Using only Numpy and Matplotlib

Primary LanguagePythonMIT LicenseMIT

DBSCAN

General description

Python implementation of 'DBSCAN' Algorithm using only Numpy and Matplotlib.

Requirements

You should install numpy and matplotlib for clustering and visualization. If you want to load sample data (sample.mat), you also need to install scipy. Although it is ironic that there is a dbscan function in the scipy... :) I've reinvented the wheel

pip3 install numpy scipy matplotlib

Usage

Initialize DBSCAN

dbscan = dbscan(x, eplison, minPts)

Run DBSCAN

idx, noise = dbscan.run()

Result Sorting

cluster, noise = dbscan.sort()

Visualize results

dbscan.plot()

Test DBSCAN

Open the runDemo file. You can just click F5 in an IDE environment to see the sample data results.

  1. sample data load

    x = io.loadmat('./sample/sample.mat')['X']

  2. Init parameters

    dbscan = DBSCAN(x,1.5,4)

  3. Run DBSCAN

    idx,noise = dbscan.run()

  4. Sorting

    g_cluster,n_cluster = dbscan.sort()

  5. Visualization

    dbscan.plot()

results

Author

Jongwon Kim : https://deep-eye.tistory.com/