Two joinable datasets on the world's Michelin three-star restaurants. The current file holds all 161 holders (restaurant, country, location, chef, year awarded, years held, plus cuisine, established year and coordinates from each restaurant's article). The panel file rebuilds the same Wikipedia list AS IT STOOD each year from 2012 through 2026 via the revision archive: 1,952 restaurant-year rows over 345 distinct restaurants, so the roughly 180 that later LOST three stars appear in their years. A snapshot mirrors what Wikipedia listed at that moment, which trails the guide's rolling announcements and includes the wiki's own early-year imperfections; it is honest for trends, not the official guide-year roster.
2,113 rows across 2 joinable files, with 17 columns in total: Restaurant, Country, Location, Chef, Awarded_Since, Years_Held, Cuisine, Established, Latitude, Longitude, Restaurant_URL, Snapshot_Year, Restaurant, Chef, Country, Location, Restaurant_URL.
It is built by programmatically scraping en.wikipedia.org, last pulled on 2026-09-03. Datasets are versioned; older versions stay downloadable.
Yes. Every CodeSights dataset is completely free as a CSV download; a free account is all it takes. Anyone can preview the data without signing in.
Yes. The exact Python scraper that built it is viewable on the dataset page by any signed-in member, so every number is reproducible.
Join the files on restaurant name; the panel's Snapshot_Year gives each restaurant's held-years run.
Automated scraping leaves room for error and the underlying sources change over time, so no version is guaranteed accurate or complete. If a number matters, verify it against the original source.