Shape and Geometry Analysis¶

Question: What are the geometric or morphological properties of features or clusters?

These methods characterise the shape and spatial configuration of polygons independently of any attribute values.

Choosing a method:

• Shape measures (esda.shape) — a suite of compactness and elongation indices (isoperimetric quotient, convex hull ratio, moment of inertia, and more). Useful for comparing the regularity of electoral districts, census units, or natural features.

• Geo-silhouettes (boundary_silhouette, path_silhouette) — spatial analogues of the silhouette score from cluster analysis. Measure how well a cluster assignment is supported by the underlying geography: a high score means observations are closer to members of their own cluster than to members of neighbouring clusters.

Shape measures operate on individual polygons; geo-silhouettes require a cluster label and a spatial graph or distance matrix.