如何在Pandas中进行Groupby Ratio计算

Question

I have a pandas dataframe as below. I need to calculate the success ratio of the Flag column when its value is equal to Y for a orgin and destination combo.

Input

ORG DSTN    FLAG
LON SIN      Y
ADL SIN      N
SIN LON      N
LON SIN      Y
LON SIN      N
ADL SIN      Y
ADL SIN      N
SIN LON      Y
SIN LON      Y
SIN LON      Y
SIN LON      N
LON SIN      N

Expected Output

ORG DSTN    FLAG    Ratio
LON SIN      Y       0.5
ADL SIN      N       0.3
SIN LON      N       0.6
LON SIN      Y       0.5
LON SIN      N       0.5
ADL SIN      Y       0.3
ADL SIN      N       0.3
SIN LON      Y       0.6
SIN LON      Y       0.6
SIN LON      Y       0.6
SIN LON      N       0.6
LON SIN      N       0.5

How can this be done in pandas.

Answer 1

Using value_counts with normalize=True :

s = (df.groupby(['ORG', 'DSTN']).FLAG
        .value_counts(normalize=True).rename('Ratio').reset_index()
)

Then changing rows where FLAG equals N to their corresponding Y value, and merging:

s.loc[s.FLAG.eq('N'), 'Ratio'] = 1.0 - s.Ratio
df.merge(s, how='left')

   ORG DSTN FLAG     Ratio
0   LON  SIN    Y  0.500000
1   ADL  SIN    N  0.333333
2   SIN  LON    N  0.600000
3   LON  SIN    Y  0.500000
4   LON  SIN    N  0.500000
5   ADL  SIN    Y  0.333333
6   ADL  SIN    N  0.333333
7   SIN  LON    Y  0.600000
8   SIN  LON    Y  0.600000
9   SIN  LON    Y  0.600000
10  SIN  LON    N  0.600000
11  LON  SIN    N  0.500000

Answer 2

well you can also group, then replace everything with the group proportion of FLAG=='Y'

 df.assign(Ratio=df.groupby(['ORG','DSTN']).FLAG.apply(lambda x:x.replace('Y|N',(x=='Y').mean(),regex=True)))
Out[174]: 
    ORG DSTN FLAG     Ratio
0   LON  SIN    Y  0.500000
1   ADL  SIN    N  0.333333
2   SIN  LON    N  0.600000
3   LON  SIN    Y  0.500000
4   LON  SIN    N  0.500000
5   ADL  SIN    Y  0.333333
6   ADL  SIN    N  0.333333
7   SIN  LON    Y  0.600000
8   SIN  LON    Y  0.600000
9   SIN  LON    Y  0.600000
10  SIN  LON    N  0.600000
11  LON  SIN    N  0.500000