I would like to see ArcGIS Pro support spatial statistical modeling directly on polyline features, not only polygons or points. This would be extremely useful for studies that focus on street networks, such as modeling traffic volume, crashes, micromobility usage, crime along streets, or infrastructure conditions.
In many urban analyses, streets are the true unit of analysis, but current spatial regression tools require converting lines into polygons or points, which can distort the underlying network structure. Ideally, spatial relationships for polylines could be defined based on network connectivity, where each street segment is considered a neighbor to other street segments it directly connects to (similar to a Queen contiguity concept for lines).
Allowing polylines to be used directly in spatial regression and spatial weights construction would greatly improve network-based urban and transportation research, reduce unnecessary preprocessing, and better reflect real-world spatial processes that operate along streets rather than areas.
