在数据框中循环功能

Question

My project: I'm creating an elo rating for tennis players, I have two different data frames.

(1)A dataframe of players with their rating (2)A dataframe of matches ordered chronologically

Working on the database of matches I would like to retrive the rating of both players and apply two functions (i already have them defined) predicted_result(rating1, rating2), and updated_rating(rating1, rating2). The first one gives me the expected result of the match given the ratings, the second one gives me the updated ratings. Finally I need to record the updated ratings in the player database.

I think that what I'm looking for is a loop that line by line:

• on the first line of the match dataframe retrives the ratings from
  the player database
• runs both functions
• replaces the old rating with the updated rating in the player database.

---

Match Dataframe

    Winner    Loser   
0   Nadal     Federer   
1   Djokovic  Verdasco   
2   Nadal     Djokovic  
3   Del Potro Verdasco 

---

Player Dataframe

    Player  Rating   
0   Nadal     2320   
1   Djokovic  2280   
2   Verdasco  2120
3   Federer   1890     
4   Del potro 1542 

I found the answer below that indicates how to roll the formula down, but I'm missing how to save the updated ratings on the player dataframe

Rolling a function on a data frame

Answer 1

The principal issue here seems to be the unhelpful format of your ratings DataFrame. Since the purpose of the index is to make it easy to access rows by index value, if you make the player name the index the problem becomes much easier. Since I don't know how ratings are calculated I have assumed that winning increases the rating by one point and losing reduces the rating by one.

First I make sure I'm using the same data as you :)

In [154]: ratings
Out[154]:
      Player  Rating
0      Nadal    2320
1   Djokovic    2280
2   Verdasco    2120
3    Federer    1890
4  Del Potro    1542

In [155]: results
Out[155]:
      Winner     Loser
0      Nadal   Federer
1   Djokovic  Verdasco
2      Nadal  Djokovic
3  Del Potro  Verdasco

Next I make a copy of the ratings table with the Player as the index.

In [156]: ir = ratings.set_index(ratings["Player"].values)

I chose to then remove the original "Player" column, since it is now redundant. YMMV.

In [157]: del ir["Player"]

In [158]: ir
Out[158]:
           Rating
Nadal        2320
Djokovic     2280
Verdasco     2120
Federer      1890
Del Potro    1542

You can iterate over each column in the results table:

In [159]: for row in results["Winner"]:
   .....:         print(row)
   .....:
Nadal
Djokovic
Nadal
Del Potro

So it's now relatively simply to update your ratings:

In [160]: for row in results["Winner"]:
   .....:         ir['Rating'][row] += 1
   .....:

In [161]: for row in results["Loser"]:
   .....:         ir['Rating'][row] -= 1
   .....:

In [162]: ir
Out[162]:
           Rating
Nadal        2322
Djokovic     2280
Verdasco     2118
Federer      1889
Del Potro    1543