**Size:** The dataset has 5 tables and 718,470 rows in total. The row counts come from the table overview (step 0 and step 4). | Table | Rows | |---|---| | raw_kaggle | 237,205 | | return_kaggle | 237,196 | | serve_kaggle | 237,185 | | players_tournament_man_ | 6,422 | | players_man_ | 462 | **Columns:** The two metadata sources disagree, so I can't give one firm figure. - The dataset card lists 78 columns: 21 in each of the three match tables, 8 in players_tournament_man_ and 7 in players_man_. - The table overview query returned 58 columns: 17, 17 and 17 for the three match tables, then 4 and 3 for the two player tables. - I did not reconcile the two counts. Running `inspect_columns` on each table would settle it. **Time period:** Yes, the data covers about 23 years. - **Match tables:** `raw_kaggle`, `return_kaggle` and `serve_kaggle` each run from **13 Aug 2001 to 28 Oct 2024**. The `Date` column is stored as text, but every row parsed as a date (for example 237,205 of 237,205 in `raw_kaggle`), so the range is reliable. - **Earlier ranges on the card:** The card showed narrower ranges, such as 2013–2024 for `raw_kaggle`. Those came from sorting the text values rather than parsing them, so they are wrong. - **players_tournament_man_:** The `year` column runs from 2002 to 2024. - **players_man_:** This table has no parsed date. Its `playing_years` column is a text list of years, and I did not check it beyond the card's listing of 2002–2024.