Missing Series name when using count() on groupby object

[andygoldschmidt]

Consider this very basic group by operation:

df = pd.DataFrame({'col1': ['a', 'b', 'a', 'c'], 'col2': [1, 3, 2, 5]})
df_grp = df.groupby('col1')['col2']

When aggregating using mean, min, ... (everything but count) the resulting Series has col2 as its name:

df_grp.mean()

col1
a       1.5
b       3.0
c       5.0
Name: col2, dtype: float64

However, if I use count for aggregation, the name is not set:

df_grp.count()

col1
a       2
b       1
c       1
dtype: int64

Not a big problem for a simple case like that, but I stumbled over that while working with a MultiIndex that needed to be reindexed and led to a KeyError due to that missing column name.

[jreback]

dup of #6124, closed by #6068

[bburan-galenea]

@jreback - Apologies for the delayed response, I have been out of the country for a few weeks. This does not appear to be a duplicate of #6124 and I verified that #6068 does not fix the problem

The problem described in this issue can be traced to SeriesGroupBy._wrap_applied_output where it just returns an (unnamed) series.

[jreback]

ok thanks
do you want to try to incorporate a fix for that issue as well ? eg add a test and fix ( in another commit), but can use same pr