Geo Cluster
Group latitude/longitude points that sit near each other on the ground — density (DBSCAN) or a latitude-compensated grid — and get a cluster id per point plus each cluster's centroid. Runs in your browser; nothing is uploaded.
Paste latitude/longitude points and choose DBSCAN or grid clustering to group nearby locations by real ground distance. The tool accepts CSV/TSV rows, JSON arrays, or GeoJSON, then returns cluster assignments and spherical centroids without sending data anywhere.
Example input:
51.5074,-0.1278,Charing Cross
51.5079,-0.1281,Nelson's Column
51.5071,-0.1274,Strand
51.5155,-0.1410,Oxford Circus
51.5161,-0.1415,Regent Street
40.7128,-74.0060,New York
With method=dbscan, radius=200, units=m, and min_points=2, the London points split into nearby clusters and the New York point is reported as unclustered noise. Switch output to CSV for spreadsheet joins, JSON for full metadata, or GeoJSON for map layers.
DBSCAN is best when clusters can have irregular shapes and outliers should stay unclustered. Grid mode is deterministic and fast for rough tiling: each latitude-compensated cell becomes a cluster when it holds enough points.
Limits: up to 10,000 points, latitude must be -90..90, longitude -180..180, and radius must be positive. CSV headers, blank lines, and # comments are ignored. GeoJSON coordinates always use longitude, latitude order; coord_order only affects CSV rows and JSON numeric pairs.
When should I use DBSCAN instead of grid?
Use DBSCAN when "nearby" means points can chain together into natural shapes, such as store visits around a neighbourhood or GPS pings along a route. Use grid when you need reproducible buckets of a fixed size and do not want clusters to cross cell boundaries.
What radius should I choose?
Pick the distance at which two points should count as the same place. For store visits or venue dedupe, try 50-200 m. For city districts, try 1-5 km. The radius is a real ground distance computed with haversine math, not a raw degree threshold.
Why are some points marked as noise?
In DBSCAN, a point is noise when it is not within radius of a core point. A core point needs at least min_points neighbours including itself. Lower min_points, increase radius, or choose grid mode if every point must receive a bucket.
Can I put the result on a map?
Yes. Set output=geojson to get a FeatureCollection containing the original points tagged with cluster ids plus centroid features for each cluster. Paste that output into a GeoJSON viewer or GIS tool.
Developer & Automation Access
Run it from the terminal
Same engine as this page, headless — via the gizza CLI:
gizza tool geo-cluster "51.5074,-0.1278,Charing Cross
51.5079,-0.1281,Nelson's Column
51.5155,-0.1410,Oxford Circus
51.5161,-0.1415,Regent Street
40.7128,-74.0060,New York"New to the CLI? Get gizza →
Open it by URL
Pre-fill and auto-run this tool with query parameters — the names match the API/CLI:
https://gizza.ai/tools/geo-cluster/?points=51.5074%2C-0.1278%2CCharing%20Cross%0A51.5079%2C-0.1281%2CNelson%27s%20Column%0A51.5155%2C-0.1410%2COxford%20Circus%0A51.5161%2C-0.1415%2CRegent%20Street%0A40.7128%2C-74.0060%2CNew%20York&method=dbscan&radius=500&units=m&min_points=2&coord_order=lat_lon&output=textMachine-readable descriptor: tool.json — title + parameters JSON Schema for agents.