Can you create new columns from a groupby object in one line using pandas?

Question

I'm wondering if I'm able to do the following in one line or if it's necessary to do it in two (I'm coming from R where I know how to do it in one call). I want to compute the batting average which requires manipulation of both the hits and the at bats columns

import pandas as pd

batting = pd.DataFrame({'playerID': [1, 1, 1, 2, 2, 2],
                        'h': [80, 97, 95, 30, 35, 22],
                        'ab': [400, 410, 390, 150, 170, 145]})

batters = (batting.groupby('playerID')
                  .agg({'h' : 'sum', 'ab' : 'sum'})
                  .reset_index())

batters['ba'] = batters['h']/batters['ab']

Do the total hits and abs matter, or just the ba column in the end? — ALollz, Jun 11 '19 at 19:33

score 4 · Accepted Answer · answered Jun 11 '19 at 19:25

eval is your friend.

(batting.groupby('playerID')
        .agg({'h' : 'sum', 'ab' : 'sum'})
        .reset_index()
        .eval('ba = h / ab'))

   playerID    h    ab        ba
0         1  272  1200  0.226667
1         2   87   465  0.187097

You can shorten this to,

batting.groupby('playerID', as_index=False).sum().eval('ba = h / ab')

   playerID    h    ab        ba
0         1  272  1200  0.226667
1         2   87   465  0.187097

Can you create new columns from a groupby object in one line using pandas?

1 Answers1