spa-l1-vienna-pip-count

Analyst notes

Description

This task probes whether the agent recognises a coverage diagnostic needs every polygon in the polygon set, not just the ones that received at least one point. The prompt deliberately does not say 'include zero-count districts' or name the spatial predicate; the persona's framing ('spot under-monitored areas') is the only cue that a left-join-back onto the full Bezirk list is required.

Approach

1. Read both layers and confirm they share a CRS, so the spatial join is meaningful.
2. Spatially join each station to the Bezirk that contains it.
3. Group by district and count the stations.
4. Left-join the counts back onto the full district list so every Bezirk appears, filling missing counts with zero.
5. Write the result as plain CSV with the three named columns, sorted by district_code.

Pitfalls

• Returning only the districts that contained at least one station, so the four zero-count Bezirke vanish and the coverage diagnostic loses its whole point.
• Pulling the wrong attribute into district_name during the join (for example the OSM relation id from the merged-in column), giving the committee integer ids where Bezirk names should be.
• Dropping the station_count column on write, leaving a two-column CSV that fails the schema gate.
• Filling every row with the same constant count instead of running the per-Bezirk join, which inflates the total far past the 49 stations actually present.
• Writing district_code under a different scheme (postal codes 1010..1230, the OSM relation id) so the values do not align with the canonical 1..23 Bezirk numbering.

Map

Recent runs task v2