# Feature Engineering for CoachAI
### Possible methods
- Categorical data
- Handling outliers
- removal
- replacing values
- capping
- discretization
-
.....

### Create features:

陳沂亨
- distance: between opponent and player
- time_diff: time(i) - time(i-1)
- classification on x-axis distance and y-axis distance
- ~~polynomial features: Do polynomial to our input feature and test the performance.~~
- shot_angle: compute the distance and the locations of two players to get the angle
- x-axis, y-axis distance
- ~~one hot encoding on shot types~~

Jean

- rank / frequency distribution of shot type of the player itself
- score_gap: roundscoreA - roundscoreB
- win ratio
- landing position variation: Compute the variation (e.g., standard deviation) of landing positions within the previous shots of the same player or rally.
- Match progress: Encode the progress of the match (e.g., 25% complete, 50% complete) as a feature.

### Question:
1. How do we know the player is A or B from the player id???

# Meta
The plot of duration for each match.
Average the duration by the number of sets.

The average duration is `22.603448275862068`. (Already averaged by #sets)

# val_given
location_area: 有一個row的area為10，不合法


### counts of location area pairs. Pairs that are not shown have value 0.
| pair | value |
|-------|-------|
| (1, 7) | 11 |
| (1, 8) | 14 |
| (2, 7) | 2 |
| (2, 8) | 20 |
| (3, 8) | 20 |
| (4, 8) | 23 |
| (5, 8) | 16 |
| (6, 8) | 16 |
| (7, 7) | 77 |
| (7, 8) | 363 |
| (8, 4) | 2 |
| (8, 5) | 3 |
| (8, 6) | 4 |
| (8, 7) | 5 |
| (8, 8) | 705 |
| (8, 9) | 14 |
| (9, 8) | 104 |

最常發生的是，兩個人都站在8號區域，也就是場地中央。


| graphs | |
| -------- | -------- |
|  | 
|  |  |
|||

## Time interval

______________________________________________________________