One row per television series Wikidata knows: 71,002 series across 183 countries, walked from the full 218-class television-series tree (dramas, sitcoms, anime, telenovelas, miniseries, reality formats, each labeled by a Type column). The columns make television measurable: premiere and end dates as ISO with year twins, a Still_Airing flag (14,779 still going), run length in years, season and episode counts with an episodes-per-season ratio, networks, genres, original language and country. The extremes are real and verified: Guiding Light's 72-year finished run leads the endings, Sandmännchen's 22,200 nightly episodes since 1959 lead the counts, and the daily soaps of India, Indonesia and Latin America legitimately air thousands of episodes in a single season, so the ranges honor them while an impossibility guard removes the true graph typos (a 16-episode K-drama claiming 16,123: even a daily soap cannot out-air the calendar at 400 episodes per year of run). Ratios are only computed where the episode count is at least the season count, because a partial count makes the ratio meaningless. A row earns its place by being usable (a real name plus an article, a premiere or an episode count); 31,933 bare shells were dropped and disclosed. Built with a 14-test acceptance suite anchored on Friends by QID (there was a 1979 sitcom of the same name), the never-ending Simpsons and Doctor Who's 1963.
71,002 rows in one CSV file, with 17 columns in total: Series, Series_ID, Type, Country, Language, Networks, Genres, Premiere_Date, Premiere_Year, End_Date, End_Year, Still_Airing, Run_Years, Seasons, Episodes, Episodes_Per_Season, Series_URL.
It is built by programmatically scraping wikidata.org, last pulled on 2026-09-14. 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.
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.