获取 pandas dataframe 中的最新值

Question

Here is a sample of my pandas dataframe

Player_A    Player_B    Gain_A    Gain_B
  John         Max        -3        3
  Max          Lucy        4       -4
  Lucy         John        1       -1
  Max          John       -5        5
  John         Lucy       -2        2

I wish to create a new column, 'Sum_2_A', which displays the sum of the most recent two instances of a player's 'Gain' (not including the value from the current row)

ie, the expected output in the given sample would be as follows

Player_A    Player_B    Gain_A    Gain_B    Sum_2_A    
  John         Max        -3        3         -3       
  Max          Lucy        4       -4          4      
  Lucy         John        1       -1          1       
  Max          John       -5        5          7         
  John         Lucy       -2        2          4         

I can do it via for loops, but it's way too slow to be useful. Any help is appreciated.

Thanks

Answer 1

IIUC, you can convert the data to long form, rolling sum on groupby:

new_df = (pd.wide_to_long(df.reset_index(), stubnames=['Player','Gain'], 
                i='index',j='type',
                sep='_', suffix = '.*'
               )
            .sort_index()
         )

new_df['Sum_2'] = (new_df.groupby('Player')
                         .Gain.rolling(3).sum()
                         .reset_index('Player',drop=True)
                         .sort_index()
                         .sub(new_df['Gain'])
                         .fillna(new_df['Gain'])
                  )

new_df.unstack('type')

Output:

      Player       Gain    Sum_2     
type       A     B    A  B     A    B
index                                
0       John   Max   -3  3  -3.0  3.0
1        Max  Lucy    4 -4   4.0 -4.0
2       Lucy  John    1 -1   1.0 -1.0
3        Max  John   -5  5   7.0 -4.0
4       John  Lucy   -2  2   4.0 -3.0