Memcached - phpCacheAdmin

PHPMem v2.0.1

Version

1.6.45

Uptime

15 days 11 hours 9 minutes 54 seconds

Memory

Total

512MB

Used

18,13MB (3.54%)

Free

493,87MB

Keys

Current

20 176

Total (since start)

33 978

Evictions

Reclaimed

377

Expired Unfetched

Evicted Unfetched

Connections

Current

6 / 1 024 max

Total

172 392

Rejected

Key

Expire (in seconds) 0 removes expiration (default).

Encode function

Value

# Three analytical questions this dataset can answer

The data holds about 237k player-match rows each for overall, serve and return stats, plus player and tournament dimensions. The tables share no join key, so each question works within one table family.

**1. How much does playing surface change a player's serve profile, and is it a real effect or a selection effect?**
- Aces per match differ sharply by surface. Clay averages about 3.4, hard about 5.9, grass about 7.6 and carpet about 7.9.
- Double faults barely move, at roughly 2.8 to 3.2 per match.
- Average serve-point winning (`SP`) is 84.1 on grass against 73.1 on clay.
- This is worth a second look because the field differs by surface. The average `Rk` on grass is about 151, against about 361 on clay and 517 on carpet. Grass matches have far more top-ranked players, so part of the "grass is a serving surface" effect may just be who plays there.
- The follow-up is to control for rank with the `*_by_Surface_Rk` gold tables. That would show whether a player of a given rank really gains on grass.

**2. How does performance change with opponent strength, and where is the upset threshold?**
- `vRk` (opponent rank) and `Rk` appear in all three fact tables. The gold tables `return_kaggle_by_Rk_against`, `return_kaggle_by_Rd_Rk` and `serve_kaggle_by_*` already slice by them.
- This lets you measure how return points won and break-point conversion degrade as opponents get stronger.
- You can also see whether the drop is gradual or a cliff, for example top-50 against the rest.
- It is the core input for win-probability or betting models, because it quantifies how much rank alone explains and how much serve and return add.

**3. Does form change by round, and do tournaments differ in how "deep" they test players?**
- `Rd` has about 14 distinct values and `Tournament` has about 3,963 distinct values (per the dataset card).
- Gold tables such as `raw_kaggle_by_Tournament_Rd` and `serve_kaggle_by_Tournment_Rd` let you compare serve and return stats from early rounds to finals.
- Comparing tournaments shows whether some events, such as grass or indoor hard-court events, favour big servers more in later rounds.
- `raw.players_tournament_man_` (6,422 rows, 2002–2024) adds career depth by player and year. It answers which tournaments and players merit coverage.

## Caveats
- `serve_kaggle` starts in 2006, but `raw_kaggle` and `return_kaggle` start in 2013, so cross-table comparisons cover different periods.
- There is a small group of 28 rows with a blank `Surface` that should be excluded.
- Carpet has only 2,477 rows, and grass has 9,479 against over 100k each for hard and clay, so carpet and grass estimates are less stable.

The chart below shows the surface differences in aces from question 1.

{"action":"final","content":"# Three analytical questions this dataset can answer\n\nThe data holds about 237k player-match rows each for overall, serve and return stats, plus player and tournament dimensions. The tables share no join key, so each question works within one table family.\n\n**1. How much does playing surface change a player's serve profile, and is it a real effect or a selection effect?**\n- Aces per match differ sharply by surface. Clay averages about 3.4, hard about 5.9, grass about 7.6 and carpet about 7.9.\n- Double faults barely move, at roughly 2.8 to 3.2 per match.\n- Average serve-point winning (`SP`) is 84.1 on grass against 73.1 on clay.\n- This is worth a second look because the field differs by surface. The average `Rk` on grass is about 151, against about 361 on clay and 517 on carpet. Grass matches have far more top-ranked players, so part of the \"grass is a serving surface\" effect may just be who plays there.\n- The follow-up is to control for rank with the `*_by_Surface_Rk` gold tables. That would show whether a player of a given rank really gains on grass.\n\n**2. How does performance change with opponent strength, and where is the upset threshold?**\n- `vRk` (opponent rank) and `Rk` appear in all three fact tables. The gold tables `return_kaggle_by_Rk_against`, `return_kaggle_by_Rd_Rk` and `serve_kaggle_by_*` already slice by them.\n- This lets you measure how return points won and break-point conversion degrade as opponents get stronger.\n- You can also see whether the drop is gradual or a cliff, for example top-50 against the rest.\n- It is the core input for win-probability or betting models, because it quantifies how much rank alone explains and how much serve and return add.\n\n**3. Does form change by round, and do tournaments differ in how \"deep\" they test players?**\n- `Rd` has about 14 distinct values and `Tournament` has about 3,963 distinct values (per the dataset card).\n- Gold tables such as `raw_kaggle_by_Tournament_Rd` and `serve_kaggle_by_Tournment_Rd` let you compare serve and return stats from early rounds to finals.\n- Comparing tournaments shows whether some events, such as grass or indoor hard-court events, favour big servers more in later rounds.\n- `raw.players_tournament_man_` (6,422 rows, 2002–2024) adds career depth by player and year. It answers which tournaments and players merit coverage.\n\n## Caveats\n- `serve_kaggle` starts in 2006, but `raw_kaggle` and `return_kaggle` start in 2013, so cross-table comparisons cover different periods.\n- There is a small group of 28 rows with a blank `Surface` that should be excluded.\n- Carpet has only 2,477 rows, and grass has 9,479 against over 100k each for hard and clay, so carpet and grass estimates are less stable.\n\nThe chart below shows the surface differences in aces from question 1.","charts":[{"kind":"bar","x":"Surface","series":["Aces__avg"],"title":"Average aces per match by surface"}]}