[FEA] Groupby with as_index=False doesn't support multiple aggregation functions

[jangorecki]

I would like to do sum(c) and mean(c) by group a. I am on 0.11.0:

import cudf as cu
import pandas as pd
pdf = pd.DataFrame({'a': [1.,3,4,1.], 'c': [6., 7., 5., 10.]})
cdf = cu.DataFrame({'a': [1.,3,4,1.], 'c': [6., 7., 5., 10.]})
ans1 = pdf.groupby(['a'],as_index=False).agg({'c': ['sum','mean']})
ans1.head()
#     a     c     
#         sum mean
#0  1.0  16.0  8.0
#1  3.0   7.0  7.0
#2  4.0   5.0  5.0
ans2 = cdf.groupby(['a'],as_index=False).agg({'c': ['sum','mean']})
#Traceback (most recent call last):
#  File "<stdin>", line 1, in <module>
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/groupby/groupby.py", line 46, in agg
#    return self._apply_aggregation(func)
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/groupby/groupby.py", line 132, in _apply_aggregation
#    result = self._groupby.compute_result(agg)
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/grou
#pby/groupby.py", line 373, in compute_result
#    return self.construct_result(out_key_columns, out_value_columns)
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/grou
#pby/groupby.py", line 460, in construct_result
#    out_value_columns, columns=self.value_names
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/grou
#pby/groupby.py", line 25, in dataframe_from_columns
#    df.columns = columns
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/data
#frame.py", line 310, in __setattr__
#    object.__setattr__(self, key, col)
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/data
#frame.py", line 1134, in columns
#    self._rename_columns(columns)
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/data
#frame.py", line 1155, in _rename_columns
#    self.rename(mapper=mapper, inplace=True)
#  File "/home/jan/anaconda3/envs/cudf/lib/python3.6/site-packages/cudf/core/data
#frame.py", line 1881, in rename
#    out_column = mapper[key] + ("cudf_" + str(postfix),)
#TypeError: must be str, not tuple

[kkraus14]

Defer this to the groupby refactor that's desperately needed.

[beckernick]

As discussed in #3429 , this error is caused by as_index=False, which we don't currently support. If you use the default as_index=True, you can do cdf.groupby(['a']).agg({'c': ['sum','mean']}).

Updating this feature request to explicitly be about supporting as_index=False.