# Numeric column distributions (partial) The evidence gives **min, max, mean and null counts** for the numeric columns. It contains **no median and no spread measure** (standard deviation, IQR, percentiles), so I can't report those. I can only say something about spread from the min–max range and from how far the mean sits within it. ## Numeric columns with min / max / mean | Table | Column | Nulls | Min | Max | Mean | |---|---|---|---|---|---| | players_man_ | number_of_matches | 0 | 1 | 577 | 20.31 | | players_tournament_man_ | year | 0 | 2002 | 2024 | 2019.74 | | raw_kaggle | Rk (player rank) | 4,388 | 1 | 2225 | 329.32 | | raw_kaggle | vRk (opponent rank) | 10,394 | 1 | 2252 | 397.45 | | raw_kaggle | TP (total points) | 86,799 | 1 | 499 | 149.01 | | raw_kaggle | SP (serve points) | 86,799 | 0 | 253 | 74.30 | | raw_kaggle | 1SP (first-serve points) | 86,799 | 0 | 165 | 45.62 | | raw_kaggle | 2SP (second-serve points) | 86,799 | 0 | 112 | 28.68 | | raw_kaggle | Aces | 86,799 | 0 | 67 | 4.93 | | raw_kaggle | DFs (double faults) | 86,799 | 0 | 26 | 2.91 | | raw_kaggle | vA (opponent aces) | 86,799 | 0 | 61 | 4.78 | | return_kaggle | Rk | 4,388 | 1 | 2225 | 329.33 | | return_kaggle | vRk | 10,394 | 1 | 2252 | 397.46 | ## What the data suggest about spread (qualitative only) - **number_of_matches** is strongly right-skewed. The mean of about 20 sits very close to the minimum of 1 and far from the maximum of 577, so a few players with very many matches pull the tail out. - **Rk and vRk** are also right-skewed. Means of about 329 and 397 sit in the low end of ranges that reach 2225 and 2252. Opponents are ranked slightly lower on average than the player. - **Aces, DFs and vA** are small-count, right-skewed variables, with means of about 5, 3 and 5 against maxima of 67, 26 and 61. - **Match-stat columns (TP, SP, 1SP, 2SP)** have means roughly a third to a half of the way up their ranges. I can't say how skewed they are without a median. - **year** runs from 2002 to 2024. The mean of 2019.7 shows most records fall in recent years. ## Caveats - **Missing data:** In raw_kaggle, the match-stat columns (TP, SP, 1SP, 2SP, Aces, DFs, vA) are null in 86,799 rows, and the profile in step 1 shows about the same count (86,793). Rk is null in 4,388 rows. Treat the means as based on the non-null subset only. I did not obtain total row counts, so I can't give the null percentage. - **Text-typed stats:** Several serve and return columns (for example DR, Dr, and the percentage columns in return_kaggle and serve_kaggle) are stored as VARCHAR, so they have no numeric summary. The serve_kaggle listing was also truncated (58 matching rows, 50 shown), so I did not see its remaining columns and any numeric ones there are not summarized. - **Not obtained:** median, standard deviation, quartiles, and distinct counts in the first output, which were blank. To finish the question, I'd need to run a query that computes `median`, `stddev` and quartiles for each column above, and to check the truncated serve_kaggle columns.