cast to float when using ``groupby.agg`` with function returning ``int`` on ``float`` input

Code Sample, a copy-pastable example if possible

In [2]: df = pd.DataFrame([[1], [2], [3.3]])

In [3]: df.groupby([1,1,1]).agg(len)
Out[3]: 
     0
1  3.0

Problem description

The result of len should be int, regardless of the input. This is not specific to len: lambda x : 3 results in the same.

Expected Output

Compare to

In [4]: df.apply(len)
Out[4]: 
0    3
dtype: int64

In [5]: df.groupby([1,1,1]).apply(len)
Out[5]: 
1    3
dtype: int64

In [6]: df.astype(int).groupby([1,1,1]).agg(len)
Out[6]: 
   0
1  3

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit: 9e7666dae3b3b10d987ce154a51c78bcee6e0728 python: 3.5.3.final.0 python-bits: 64 OS: Linux OS-release: 4.9.0-3-amd64 machine: x86_64 processor: byteorder: little LC_ALL: None LANG: it_IT.UTF-8 LOCALE: it_IT.UTF-8

pandas: 0.21.0.dev+265.g9e7666dae pytest: 3.0.6 pip: 9.0.1 setuptools: None Cython: 0.25.2 numpy: 1.12.1 scipy: 0.19.0 xarray: None IPython: 5.1.0.dev sphinx: 1.5.6 patsy: 0.4.1 dateutil: 2.6.0 pytz: 2017.2 blosc: None bottleneck: 1.2.1 tables: 3.3.0 numexpr: 2.6.1 feather: 0.3.1 matplotlib: 2.0.2 openpyxl: None xlrd: 1.0.0 xlwt: 1.1.2 xlsxwriter: 0.9.6 lxml: None bs4: 4.5.3 html5lib: 0.999999999 sqlalchemy: 1.0.15 pymysql: None psycopg2: None jinja2: 2.9.6 s3fs: None pandas_gbq: None pandas_datareader: 0.2.1

[toobaz]

you may have integers as a result, so you can try to downcast

My point was that the result of len is an int, and should never become a float, so there should be no need to downcast (the fact that input data was float should be irrelevant). But I will probably just need to look at the code to understand what you mean.

[rhshadrach]

Edit: The following comment is not relevant to the issue here since the handling of a string argument differs from that of a callable.

I ran into a similar issue with .transform(‘nunique’):

df = pd.DataFrame([[1, 1.1], [1, 3.1], [2, 1.1]], columns=['a', 'b'])
df.groupby('a').b.transform('nunique')

The resulting series I get are floats using 0.25.3. They become integers if column b values are integers, or if I replace .transform(‘nunique’) with merely .nunique().