groupby error when grouping by float index with non-unique values

[jeybrahms]

Code Sample, a copy-pastable example if possible

pd.Series([2, 5, 6, 8], [2.0, 4.0, 4.0, 5.0]).groupby(level=0)

Problem description

This should just work instead of raising.

Expected Output

Output of pd.show_versions()

pandas : 0.26.0.dev0+593.g9d45934af
numpy : 1.17.0

[WillAyd]

Interesting find. I think this stems from some clean up in #27909 but very surprising that we would throw anything other than a Key or Index error at the below line

pandas/pandas/core/groupby/grouper.py

Line 602 in ee59b7d

except (KeyError, IndexError):

@jorisvandenbossche @jbrockmendel

[WillAyd]

Traceback on master for reference

>>> pd.Series([2, 5, 6, 8], [2.0, 4.0, 4.0, 5.0]).groupby(level=0)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/williamayd/clones/pandas/pandas/core/generic.py", line 7943, in groupby
    **kwargs
  File "/Users/williamayd/clones/pandas/pandas/core/groupby/groupby.py", line 2494, in groupby
    return klass(obj, by, **kwds)
  File "/Users/williamayd/clones/pandas/pandas/core/groupby/groupby.py", line 391, in __init__
    mutated=self.mutated,
  File "/Users/williamayd/clones/pandas/pandas/core/groupby/grouper.py", line 607, in _get_grouper
    if is_in_obj(gpr):  # df.groupby(df['name'])
  File "/Users/williamayd/clones/pandas/pandas/core/groupby/grouper.py", line 601, in is_in_obj
    return gpr is obj[gpr.name]
  File "/Users/williamayd/clones/pandas/pandas/core/series.py", line 1088, in __getitem__
    result = self.index.get_value(self, key)
  File "/Users/williamayd/clones/pandas/pandas/core/indexes/numeric.py", line 420, in get_value
    loc = self.get_loc(k)
  File "/Users/williamayd/clones/pandas/pandas/core/indexes/numeric.py", line 479, in get_loc
    return super().get_loc(key, method=method, tolerance=tolerance)
  File "/Users/williamayd/clones/pandas/pandas/core/indexes/base.py", line 2826, in get_loc
    return self._engine.get_loc(key)
  File "pandas/_libs/index.pyx", line 106, in pandas._libs.index.IndexEngine.get_loc
    cpdef get_loc(self, object val):
  File "pandas/_libs/index.pyx", line 128, in pandas._libs.index.IndexEngine.get_loc
    return self._get_loc_duplicates(val)
  File "pandas/_libs/index.pyx", line 143, in pandas._libs.index.IndexEngine._get_loc_duplicates
    left = values.searchsorted(val, side='left')
TypeError: '<' not supported between instances of 'NoneType' and 'NoneType'

[WillAyd]

@jbrockmendel do you mind taking a look at this one? I think would be good to fix before releasing 1.0

[jbrockmendel]

Sure, probably next week.

[WillAyd]

Awesome thanks!

…

Sent from my iPhone

On Nov 14, 2019, at 6:28 PM, jbrockmendel ***@***.***> wrote:  Sure, probably next week. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.