← Back to dataset

Dataset methodology report

Generated September 23, 2026 at 20:16:07 UTC
Download report.mdDownload CSV
Result

Methodology

  1. OpenStreetMap was used only as a discovery inventory. Physical restaurant locations were manually curated, and discovery links were replaced with official restaurant sites or official brand location pages.
  2. Every official URL was requested live. Status, redirect target, title, content hash, and lightweight address/name checks were recorded in source_checks.json. A 401, 403, 405, or 429 is retained as reachable-but-automation-blocked rather than treated as a successful content read.
  3. Every complete address was geocoded with the public Esri World Geocoding Service. The selected candidate had to score at least 75 and fall in the district polygon. Requests and candidate responses are preserved in geocode_evidence.json.
  4. The U.S. Census Bureau 2025 TIGER/Line unified school district shapefile was loaded and Troy School District, Michigan, GEOID 2634260, was selected. Polygon covers was used so points exactly on the legal boundary pass.
  5. Addresses were independently geocoded with the City of Troy Site Address Locator and checked against the City's official Administrative/MapServer/3 feature NAME='TROY SCHOOLS'.
  6. The clean verifier enforced the exact seven-column CSV schema, 51 rows, nonblank cells, permitted likelihood values, URL syntax, unique normalized addresses, and both boundary passes.

Sources

Verification outcomes

Caveats