About this dataset

DOHMH's Restaurant Inspection Results, mirrored on NYC Open Data as Socrata dataset 43nn-pn8j. Each row is one sustained violation issued during one inspection — so a single inspection that cited four violations produces four rows here. The grade card hung in the restaurant window comes from the most recent inspection's grade field.

Source

Catalog page: NYC Open Data
Endpoint (SODA v3): POST https://data.cityofnewyork.us/api/v3/views/43nn-pn8j/query.json
Rows: — (one per sustained violation per inspection)

The 1900-01-01 problem

The dataset includes restaurants that have filed paperwork to operate but have not yet been physically inspected. These rows are assigned a placeholder inspection_date of 1900-01-01T00:00:00.000. Any temporal analysis that doesn't filter these out gets dragged into the early 1900s — a famous gotcha that's burned both data journalists and ML practitioners. Every VIEW template in our adapter automatically applies inspection_date > '2000-01-01' as a baseline filter.

Caveats

Cuisine descriptions are inspector-categorized, not restaurant-self-reported. The category "American" is enormous and absorbs anything not obviously another cuisine.
Borough strings here are title-case ("Manhattan", "Brooklyn"), not uppercase like NYC 311's borough column. Don't reuse a join key without checking case.
A single restaurant (camis) can appear with different display names if it changed ownership; the camis stays stable, the dba drifts.

Citation

NYC Department of Health and Mental Hygiene (2026). Restaurant Inspection Results. Retrieved recently via NYC Open Data SODA v3.