Use case 3

3️⃣ Complex Spatial Join (Proximity)

• Goal: To test performance on computationally intensive join operations on large datasets.

• Example task: Given the road network and buildings in Rome, count for each main road (highway = 'primary') how many buildings are within 20 meters.

• Workflows to compare:

1. DuckDB: The query with ST_DWithin that I have already analyzed and seen to be slow. I should also discuss the optimization strategies I found (e.g. batching).

2. PostGIS: The same query, but it will transparently use spatial indexes (to be created first).

3. GeoPandas: gpd.sjoin_nearest(roads, buildings, max_distance=20)

• Metrics: Execution time, correctness of results, complexity of the code.