How to calculate average percentages of values within group?-CodePudding

I have a dataframe:

date             type
2021-08-12       fail
2021-08-12       fail
2021-08-12       win
2021-08-12       great_win
2021-08-13       fail
2021-08-13       win
2021-08-13       win
2021-08-13       win

I want to calculate percentage of each 'type' within date group and then average values among all dates. So desired results must be:

date             type          type_perc
2021-08-12       fail           0.5
2021-08-12       win            0.25
2021-08-12       great_win      0.25
2021-08-13       fail           0.25
2021-08-13       win            0.75
2021-08-13       great_win      0.0

and then average among all dates. this is the desired final result:

type          type_perc
fail           0.375
win            0.5
great_win      0.175

How to do that?

CodePudding user response：

You can try this:

tmp = df.groupby(['date', 'type']).size()/df.groupby('date')['type'].size()
print(tmp)

date        type
2021-08-12  fail         0.50
            great_win    0.25
            win          0.25
2021-08-13  fail         0.25
            win          0.75
dtype: float64

result = tmp.groupby(level=1).sum()/tmp.sum()
print(result)

type
fail         0.375
great_win    0.125
win          0.500
dtype: float64

or this:

result = tmp.groupby(level=1).mean()
print(result)
type
fail         0.375
great_win    0.250
win          0.500
dtype: float64

It's not quite clear by your question