Meeting notes 24-02-05

[anastassiavybornova]

concrete next steps

• everyone continues working on their simplification workflow, aim is to be able to evaluate (see below) & compare them numerically, and to point out where the respective issues are
• make sure to use the same input data (/data/)
• if someone has a lot of time > next one up is Robin's tesselation method (cf issues)
• Martin's students will hopefully soon start manual simplification which will give us the ground truth 🤞

thoughts re evaluation techniques

• number of artifacts (use always the threshold from the BASE network)
• total area of artifacts
• node degree histogram (high-degree nodes are an indicator of trouble , cf. "spaghetti")
• for each city, specific pre-defined locations (plot and see what happens)
• todo here: write an evaluation function (or a set thereof)? that we all can use func(gdf, params) --> gdf, eval_metrics

ideas to explore further down the line:

• simplification should be location-based (a): take osmnx.simplification.consolidate_intersections and integrate an additional check in it - only add to cluster if face artifact is present at that location (tbd how to exactly define this)
• simplification should be location-based (b): take face artifacts of original street network (momepy.FaceArtifacts) and tesselate them, potentially with Robin's method? (cf also Martin's pre-bananas work)
• dbscan-like algorithm to identify local node densities