DBSCAN Clustering

A density-based algorithm for discovering clusters of arbitrary shape and isolating noise.

How it works

Step 1: Neighborhoods

Draw the Epsilon (ε) search radius around every point.

Step 2: Density

If a point has ≥ MinPts in its radius, it is a Core. Else, it is Noise.

Step 3: Connect

Connect overlapping Core Points to form cluster groups.

Configuration

Phase: Initialization Clusters: 0

Hyperparameter Grid Search

Green indicates 3 Clusters found. Bold borders lock on success.