pandas groupby and transform only one column

Asked Feb 08 '20 at 14:36

Active Feb 08 '20 at 15:16

Viewed 29 times

I'd like to groupby and transform only one of the grouped columns.

Suppose I have dataframe like this:

dataf = pd.DataFrame({'question': ['q1', 'q1', 'q1', 'q2', 'q2', 'q2', 'q3', 'q3', 'q3']*10, 'value': [1,2,3,1,2,3,1,2,3]*10})
datafG = dataf.groupby(['question', 'value']).agg({'value': 'count'})
datafG['sum_n'] = datafG.groupby(['question']).transform('sum')

And then I want to groupby and get proportions of the count column datafG / datafG.groupby(level=[0]).transform('sum') this will however transform both, I am looking for a way to keep my sum_ncolumn as it is.

This gives me a df looking like this:

                    value   sum_n
question    value   
q1          1       0.33    0.33
q1          2       0.33    0.33
q1          3       0.33    0.33
q2          1       0.33    0.33
q2          2       0.33    0.33
q2          3       0.33    0.33
q3          1       0.33    0.33
q3          2       0.33    0.33
q3          3       0.33    0.33

While I want:

                    value   sum_n
question    value   
q1          1       0.33    30
q1          2       0.33    30
q1          3       0.33    30
q2          1       0.33    30
q2          2       0.33    30
q2          3       0.33    30
q3          1       0.33    30
q3          2       0.33    30
q3          3       0.33    30

edited Feb 08 '20 at 15:16

asked Feb 08 '20 at 14:36

honetkri

add a dataframe showing your expected output. – sammywemmy Feb 08 '20 at 14:53
```dataf.groupby(...)[column_u_want_to_agg].count()``` ? – Grzegorz Skibinski Feb 08 '20 at 15:04
The outcome you are getting now and the outcome you want to get would be helpful! – coffeinjunky Feb 08 '20 at 15:05
But it sounds like this might help? https://stackoverflow.com/questions/23377108/pandas-percentage-of-total-with-groupby – coffeinjunky Feb 08 '20 at 15:06
looks like you need `datafG['sum_n'] = datafG.groupby(level=0)['value'].transform('sum')` ? – anky Feb 08 '20 at 15:07
I have updated with expected output – honetkri Feb 08 '20 at 15:16

pandas groupby and transform only one column

0 Answers0