Groupby as columns with MultiIndex

Question

I have a dataframe similar to

df = pd.DataFrame(
    pd.np.random.randint(1, 10, (20, 2)),
    index=["a", "b", "c", "d"] * 5,
    columns=["foo", "bar"],
)

looking something like this

I'm trying to get it into the following multi-index form: two columns for each of the 4 axis labels each with a "foo" and "bar" sub-column:

I've tried fiddling around with df.groupby(df.index) but then I can't figure out how to turn the groups into columns.

Answer 1

Use GroupBy.cumcount for MultiIndex in index with DataFrame.set_index , then reshape by DataFrame.unstack and change order of Multiindex in columns by DataFrame.swaplevel with DataFrame.sort_index :

df1 = (df.set_index(df.groupby(level=0).cumcount(), append=True)
         .unstack(0)
         .swaplevel(0,1, axis=1)
         .sort_index(axis=1, level=[0,1], ascending=[True, False]))
print (df1)
    a       b       c       d    
  foo bar foo bar foo bar foo bar
0   6   8   1   6   8   4   6   1
1   2   6   4   5   3   1   5   4
2   2   5   2   9   5   4   8   9
3   5   2   6   6   7   3   2   9
4   4   8   9   6   4   2   1   5