Netflix Catalog Csv Snapshot Data: Dataset Structure and Field Coverage

A netflix catalog csv snapshot collapses a streaming platform's catalog into one flat table where every row is a title. The version analysts actually use is Kaggle - Netflix Movies and TV Shows, published by Shivam Bansal as shivamb/netflix-shows and packaged as a single netflix_titles.csv. Version 5, its final release, landed on September 27, 2021, holding 8,807 data rows at roughly 3.4 MB uncompressed and 1.4 MB compressed.

Twelve columns cover identification, people, geography, timing, classification and synopsis. show_id keys each row sequentially from s1, and type splits the catalog into Movie and TV Show. Timing arrives twice: release_year records the original release - titles run back to 1925 - while date_added, formatted like September 25, 2021, records when Netflix listed the title.

The table below maps every column to its role.

Two quirks matter when parsing. duration mixes units, carrying '90 min' strings for movies and '2 Seasons' counts for TV shows, so split on the suffix before casting anything numeric. director is empty for most TV shows, and country can be blank, so any geography-based join needs a fallback.

The opening rows show the pattern. s1 is Dick Johnson Is Dead, a 90-minute United States documentary rated PG-13 and added on September 25, 2021. s2 is Blood & Water, a South African TV show running 2 Seasons, rated TV-MA and filed under International TV Shows, TV Dramas and TV Mysteries.

Pair the frozen panel with sources that still move. Three routes cover the gap.

Programmatic refresh: The Movie Database (TMDB) API v3 serves title detail and watch providers under a free key at roughly 40 requests per second, and publishes daily gzipped JSONL ID exports for seven media types; the export job runs around 07:00 UTC, files land by 08:00 UTC and are retained for three months.

For platform comparisons, the same Kaggle author ships matching Amazon Prime Video, Disney+ and Hulu catalog datasets, so the 8,807-row Netflix table drops straight into a four-platform content-mix study. Preference modeling wants ratings instead: Kaggle's The Movies Dataset pairs 45,000 films with 26 million ratings from 270,000 users under commercial delivery terms, though its releases stop at July 2017.

Treat the snapshot as one layer of a stack: a dated catalog table for content-mix history, a daily availability source for the present, and a ratings corpus for demand. The movies entertainment data guide walks the whole industry - 15 pooled datasets spanning title metadata, audience ratings, box-office revenue and streaming availability - and maps which combination fits recommender benchmarks, SVOD strategy reviews or box-office comparisons.