The dataset has 5 tables with **718,470 rows** in total, and the match data covers **13 Aug 2001 to 28 Oct 2024**. **Rows** | Table | Rows | |---|---| | raw_kaggle | 237,205 | | return_kaggle | 237,196 | | serve_kaggle | 237,185 | | players_tournament_man_ | 6,422 | | players_man_ | 462 | The row counts come from the table overview query. The three match-level tables are close in size but not identical, and there is no join key between them. **Columns** The two sources I have disagree, so I can't give a single figure with confidence: - The table overview query reports 17 + 17 + 17 + 4 + 3 = **58 columns**. - The dataset card lists 21 + 21 + 21 + 8 + 7 = **78 columns**. The gap is probably a difference in what each source counts, but I didn't run anything to confirm that. Calling `inspect_columns` on each table would settle it. **Time period** - **Match tables:** the `Date` column is stored as text, so I parsed it into real dates. All rows in raw_kaggle, return_kaggle and serve_kaggle parsed, and each table runs from **2001-08-13 to 2024-10-28**. - **Earlier ranges on the card:** the card's ranges, such as "1-Apr-2013" for raw_kaggle and "10-Apr-2006" for serve_kaggle, come from sorting the text values alphabetically. They are not real earliest and latest dates, so ignore them. - **players_tournament_man_:** the `year` column runs from **2002 to 2024**. - **players_man_:** `playing_years` is a text list of years, so I didn't compute a range from it. The card shows years between 2002 and 2024.