The Madrid Housing Squeeze — a federated data story

A data story about housing in Madrid, told from two federated Portolan catalogs queried in place with DuckDB — no downloads, no server, every figure re-runnable with its provenance. Built in the spirit of the OGC Connect Helsinki demo (federate real publishers; keep the publisher + query on every number).

Thesis: demand-driven prices are back above the 2007 peak while new supply runs at a quarter of its mid-2000s rate — and within the city the rent burden concentrates in the centre. See FINDINGS.md.

• Comunidad de Madrid — https://storage.googleapis.com/carto-portolan-madrid/comunidad-madrid — regional housing market (prices, permits, rents, stock).
• City of Madrid — https://storage.googleapis.com/carto-portolan-madrid/madrid-city — within-city geometry & layers.

python3 tools/analyze.py     # federated DuckDB queries over both catalogs → data/*.json + data/*.geojson
python3 tools/fetch_vut.py   # city zoom: licensed tourist dwellings (VUT) by district

Every query is in QUERIES.md; every output JSON carries a _provenance block.

Analysis phase complete (deepened) — 9 figures validated in data/: affordability map, the squeeze (price vs permits), affordability ratio (×income), city rent by postal code, rent index by district, supply mix, and the VUT tourist-flat city zoom. Presentation (scrollytelling page vs full Helsinki-style federation webapp) is the next step — see FINDINGS.md for the open items.

FINDINGS.md        the story + headline numbers + provenance + honesty notes
QUERIES.md         every figure's DuckDB query
tools/analyze.py   federated analysis → data/
data/*.json|geojson  figure-ready outputs (git-tracked; small)