This skill should be used when the user asks to "find clusters in this data", "are these points clustered or random", "show me hot spots", "kernel density of features", "spatial clustering with DBSCAN", "find areas with high point density", "nearest neighbor analysis", "spatial pattern detection", or invokes `/gis-to-db:analyze-patterns`. Runs spatial pattern analysis on a GIS point layer: DBSCAN clustering, nearest-neighbor distance distribution, kernel density estimation summary, and convex hulls of detected clusters. Returns labeled features (which cluster each point belongs to) plus a summary report.
This skill should be used when the user asks to "find clusters in this data", "are these points clustered or random", "show me hot spots", "kernel density of features", "spatial clustering with DBSCAN", "find areas with high point density", "nearest neighbor analysis", "spatial pattern detection", or invokes `/gis-to-db:analyze-patterns`. Runs spatial pattern analysis on a GIS point layer: DBSCAN clustering, nearest-neighbor distance distribution, kernel density estimation summary, and convex hulls of detected clusters. Returns labeled features (which cluster each point belongs to) plus a summary report.
Spatial Pattern Analysis
v0.1 status — production-ready for point layers. Polygon/line clustering deferred to v0.2.
Clustering — DBSCAN result: number of clusters found, noise-point count, per-cluster size + centroid + convex hull area.
Nearest-neighbor analysis — mean / median / std nearest-neighbor distance; ratio vs expected for random distribution (Clark-Evans R: <1 = clustered, ≈1 = random, >1 = dispersed).
Recommendations — when to adjust --eps (cluster radius) or --min-samples based on output.
Plus an output GeoJSON (--out-file) with original points labeled by cluster ID (-1 = noise).
