Point cloud (top) · colored by extracted cluster valley Reachability plot (bottom) · bar height = reachability-distance

OPTICS: Reachability Plot & Varying-Density Clustering (2D)

OPTICS is the density-based clustering algorithm that fixes DBSCAN's biggest weakness: a single global radius that cannot fit clusters of very different densities at once. This 2D simulator generates a point cloud with deliberately uneven density — a tight cluster, a medium cluster, a sparse cluster and scattered noise — then animates the real OPTICS priority-queue traversal on it: the current point is highlighted as it is popped from the seed queue, and the reachability plot grows live underneath, one bar per visit, in the algorithm's own order. Adjust ε and MinPts to change what counts as a neighborhood and a core point, then slide the extraction threshold to watch the valleys in the reachability plot merge, split and turn to noise — repainting the point cloud above directly from those valleys, the same interactive read a data scientist does on a real OPTICS reachability plot before picking a cut.