Home › Methodology

Methodology

1. Download. The full public datasets are downloaded from each city's official open data portal.
2. Group. Rows are grouped by the city's own identifier: the CAMIS number in New York City, the license number in Chicago. New York publishes one row per violation; these are regrouped into one inspection per date and inspection type.
3. Keep the city's terms. Scores, grades, results, inspection types, violation codes and descriptions are shown as published. No national score is calculated.
4. Check. Records without a usable date or identifier are skipped. Placeholder dates (New York uses 01/01/1900 for establishments not yet inspected) are not treated as inspections.
5. Summarize carefully. Counts and medians on this site are calculated directly from the imported records. Nothing is estimated or filled in.
6. Link back. Every establishment page links to its official record so you can check it.

Text formatting

Names and addresses are published in capital letters by both cities; this site displays them in title case for readability. Long Chicago inspector comments are shortened; the full text is in the official record.

Which restaurants have their own page

All establishments can be found with search and appear on their ZIP code pages. At launch, a sample of establishments with at least two inspections also has a standalone page; more will be added over time. Selection is by a neutral, fixed method that does not consider grades, results or violations.

Limits

• Inspection results describe one visit. They can change quickly.
• Cities sometimes correct or remove records. Our copy may lag behind.
• Some establishments appear under a former name or ownership.