Density-Based Clustering
DBSCAN
A Practical Field Guide
A focused, independent resource for density-based clustering. Learn how DBSCAN finds arbitrary-shaped clusters, labels noise, chooses epsilon and minPts, and compares with related methods.
Concept
Core points, border points, noise, density reachability, and cluster shape.
Math
Distance metrics, complexity, validation, and high-dimensional behaviour.
Code
Practical Python and R guides, parameter selection, and spatial indexing.
Compare
DBSCAN versus k-means, HDBSCAN, GMM, and hierarchical clustering.
Papers
Annotated readings of Ester et al. 1996 and later density-based research.
Latest entries
DBSCAN Benchmark Datasets · sklearn Clustering Datasets and What Each One Tests
make_moons, make_circles, make_blobs, anisotropic and the null case: what each sklearn toy dataset exposes in DBSCAN, with measured results.
Density Reachability · Core Points, Border Points and the ε-Neighbourhood Definition
The eps-neighbourhood, core points, border points and noise written out precisely, with a replaceable distance function and a counted example.
Feature Scaling Before DBSCAN · Why Standardisation Changes Your Neighbourhoods
DBSCAN's eps is a radius in your data's own units, so unit choice decides which points count as neighbours. Measured numbers and the wiring.
DBSCAN vs Streaming Clustering · Batch Density Clustering Against Online Methods
DBSCAN has no incremental method at all. What BIRCH's CF-tree and MiniBatchKMeans do instead, and why OPTICS is not the answer.
DBSCAN for Geospatial Data · Clustering GPS Coordinates
How to cluster geographic data with DBSCAN — Haversine distance, handling latitude and longitude, map visualisation with folium, and real-world spatial analysis.
Scaling DBSCAN to Large Datasets · Approximation Strategies
Practical approaches to clustering millions of points — approximate DBSCAN, sampling, grid-based acceleration, and the limits of exact DBSCAN.