BUG: SeriesGroupBy.transform should raise with axis=1

[arw2019]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Code Sample, a copy-pastable example

The dropna=True case runs

In [7]: df = pd.DataFrame({"A": [0, 0, 7, 1], "B": [1, 2, 4, 3]}, index=list('abcd')) 
   ...: gb = df.groupby('B', dropna=True, axis=1) 
   ...: gb['B'].transform(len)                                                                         
Out[7]: Series([], Name: B, dtype: int64)

but with dropna=False the call to transform throws with a somewhat unfriendly message

In [9]: df = pd.DataFrame({"A": [0, 0, 7, 1], "B": [1, 2, 4, 3]}, index=list('abcd')) 
   ...: gb = df.groupby('B', dropna=False, axis=1) 
   ...: gb['B'].transform(len)                    
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-2-c7271aa13f24> in <module>
      1 df = pd.DataFrame({"A": [0, 0, 7, 1], "B": [1, 2, 4, 3]}, index=list('abcd'))
      2 gb = df.groupby('B', dropna=False, axis=1)
----> 3 gb['B'].transform(len)

/workspaces/pandas-arw2019/pandas/core/groupby/generic.py in transform(self, func, engine, engine_kwargs, *args, **kwargs)
    506 
    507         if not isinstance(func, str):
--> 508             return self._transform_general(func, *args, **kwargs)
    509 
    510         elif func not in base.transform_kernel_allowlist:

/workspaces/pandas-arw2019/pandas/core/groupby/generic.py in _transform_general(self, func, *args, **kwargs)
    546 
    547             concatenated = concat(results)
--> 548             result = self._set_result_index_ordered(concatenated)
    549         else:
    550             result = self.obj._constructor(dtype=np.float64)

/workspaces/pandas-arw2019/pandas/core/groupby/groupby.py in _set_result_index_ordered(self, result)
    691             result = result.sort_index(axis=self.axis)
    692 
--> 693         result.set_axis(self.obj._get_axis(self.axis), axis=self.axis, inplace=True)
    694         return result
    695 

/workspaces/pandas-arw2019/pandas/core/series.py in set_axis(self, labels, axis, inplace)
   4365     @Appender(generic.NDFrame.set_axis.__doc__)
   4366     def set_axis(self, labels, axis: Axis = 0, inplace: bool = False):
-> 4367         return super().set_axis(labels, axis=axis, inplace=inplace)
   4368 
   4369     @doc(

/workspaces/pandas-arw2019/pandas/core/generic.py in set_axis(self, labels, axis, inplace)
    657         """
    658         self._check_inplace_and_allows_duplicate_labels(inplace)
--> 659         return self._set_axis_nocheck(labels, axis, inplace)
    660 
    661     def _set_axis_nocheck(self, labels, axis: Axis, inplace: bool):

/workspaces/pandas-arw2019/pandas/core/generic.py in _set_axis_nocheck(self, labels, axis, inplace)
    662         # NDFrame.rename with inplace=False calls set_axis(inplace=True) on a copy.
    663         if inplace:
--> 664             setattr(self, self._get_axis_name(axis), labels)
    665         else:
    666             obj = self.copy()

/workspaces/pandas-arw2019/pandas/core/generic.py in __setattr__(self, name, value)
   5384         try:
   5385             object.__getattribute__(self, name)
-> 5386             return object.__setattr__(self, name, value)
   5387         except AttributeError:
   5388             pass

/workspaces/pandas-arw2019/pandas/_libs/properties.pyx in pandas._libs.properties.AxisProperty.__set__()
     64 
     65     def __set__(self, obj, value):
---> 66         obj._set_axis(self.axis, value)

/workspaces/pandas-arw2019/pandas/core/series.py in _set_axis(self, axis, labels, fastpath)
    425         if not fastpath:
    426             # The ensure_index call above ensures we have an Index object
--> 427             self._mgr.set_axis(axis, labels)
    428 
    429     # ndarray compatibility

/workspaces/pandas-arw2019/pandas/core/internals/managers.py in set_axis(self, axis, new_labels)
    217 
    218         if new_len != old_len:
--> 219             raise ValueError(
    220                 f"Length mismatch: Expected axis has {old_len} elements, new "
    221                 f"values have {new_len} elements"

ValueError: Length mismatch: Expected axis has 2 elements, new values have 4 elements
                                                     

Expected Output

Since there are no missing values in the input dropna=True and dropna=False should give the same results.

Problem description

(Edited)
As discussed below and in #35751 SeriesGroupBy with axis=1 never makes sense. We should have a more explicit error message.

Output of pd.show_versions()

INSTALLED VERSIONS

commit : 0cdea22
python : 3.8.3.final.0
python-bits : 64
OS : Linux
OS-release : 5.4.0-47-generic
Version : #51-Ubuntu SMP Fri Sep 4 19:50:52 UTC 2020
machine : x86_64
processor :
byteorder : little
LC_ALL : C.UTF-8
LANG : C.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.2.0.dev0+354.g0cdea2261
numpy : 1.18.5
pytz : 2020.1
dateutil : 2.8.1
pip : 20.1.1
setuptools : 49.1.0.post20200704
Cython : 0.29.21
pytest : 5.4.3
hypothesis : 5.19.0
sphinx : 3.1.1
blosc : None
feather : None
xlsxwriter : 1.2.9
lxml.etree : 4.5.2
html5lib : 1.1
pymysql : None
psycopg2 : 2.8.5 (dt dec pq3 ext lo64)
jinja2 : 2.11.2
IPython : 7.16.1
pandas_datareader: None
bs4 : 4.9.1
bottleneck : 1.3.2
fsspec : 0.7.4
fastparquet : 0.4.0
gcsfs : 0.6.2
matplotlib : 3.2.2
numexpr : 2.7.1
odfpy : None
openpyxl : 3.0.4
pandas_gbq : None
pyarrow : 0.17.1
pytables : None
pyxlsb : None
s3fs : 0.4.2
scipy : 1.5.0
sqlalchemy : 1.3.18
tables : 3.6.1
tabulate : 0.8.7
xarray : 0.15.1
xlrd : 1.2.0
xlwt : 1.3.0
numba : 0.50.1

[arw2019]

take

[rhshadrach]

I think allowing axis=1 for a Series is largely (entirely?) not implemented. Doing pd.Series([1, 2, 3]).groupby([0], axis=1) raises, and I can't find any way of making something like:

df = pd.DataFrame({"A": [0, 0, 7, 1]})
df.groupby([0], axis=1)['A'].sum()

work. I think this "backdoor" to getting a SeriesGroupBy with axis=1 is a bug. See #35443. I'm thinking the only case where it maybe makes sense to do this this when as_index=False.

[arw2019]

I think I agree that it's a bug, except possibly the as_index=False case.

TODOs:

• should we add tests for this?
• do you think it's a good idea to throw specific error messages when a user tries this? If yes I'm happy to take a stab at that and see how complicated it gets / if it's worth the effort

[rhshadrach]

@arw2019 - Looks to me like this was closed along with #37725 (which I think should have been marked as a duplicate of this). Do you agree?

[arw2019]

Yes, agreed. Duplicate of the other issue and can be closed.

[rhshadrach]

Thanks @arw2019