Pandas how to flatten columns after agg function?

Question

say I have a df:

data=[('a', 1), ('a', 1),('b', 1),('a', 3),('b', 2),('c', 1),('a', 2),('b', 3),('a', 2)]
df=df=pd.DataFrame(data, columns=['project', 'duration'])

# Then I made an aggregation:
df_agg=df.groupby('project').agg({'duration': ['median', 'mean']}).reset_index()

Out[11]: 
  project duration     
            median mean
0       a        2  1.8
1       b        2  2.0
2       c        1  1.0


In [12]: df_agg.info()                                                                                                                                                    
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 3 entries, 0 to 2
Data columns (total 3 columns):
(project, )           3 non-null object
(duration, median)    3 non-null int64
(duration, mean)      3 non-null float64
dtypes: float64(1), int64(1), object(1)
memory usage: 152.0+ bytes

However, the df_agg is not like an ordinary DataFrame, because the columns look like a tuple (duration, median) , so that I can't get the columns conveniently with df[['median', 'mean']]

My question is how can I change the df_agg to an ordinary DataFrame, with the columns flattened?

Answer 1

The df_agg dataframe has a MultiIndex for its columns. Only this has to be flattened.

A trivial way is to convert it to a list and join each element:

df_agg.columns = ['_'.join(col) for col in df_agg.columns]

it gives:

  project_  duration_median  duration_mean
0        a                2            1.8
1        b                2            2.0
2        c                1            1.0

If you want you can then rename the columns to have nicer names

Answer 2

you could perform the aggregation on the selected column for the groupby:

df.groupby('project')['duration'].agg(['median', 'mean']).add_prefix('duration_').reset_index()

• Thank you Chris A for the add_prefix before reset_index idea

output:

    project duration_median duration_mean
0   a       2               1.8
1   b       2               2.0
2   c       1               1.0