BUG: cython version of groupby.cummax throws error on datetimes

[adbull]

Code Sample, a copy-pastable example if possible

>>> import pandas as pd
>>> x = pd.DataFrame(dict(a=[1], b=pd.to_datetime(['2001'])))
>>> x.groupby('a').b.cummax()

Traceback (most recent call last):
  File "<ipython-input-9-316257648d5f>", line 1, in <module>
    x.groupby('a').b.cummax()
  File "~/anaconda3/lib/python3.5/site-packages/pandas/core/groupby.py", line 1454, in cummax
    return self._cython_transform('cummax', **kwargs)
  File "~/anaconda3/lib/python3.5/site-packages/pandas/core/groupby.py", line 806, in _cython_transform
    raise DataError('No numeric types to aggregate')
DataError: No numeric types to aggregate

Problem description

The current github version of pandas has cython implementations of groupby.cummin and groupby.cummax, which throw an error if called on datetime columns. (See #15048, 0fe491d.)

Expected Output

0   2001-01-01
Name: b, dtype: datetime64[ns]

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.5.2.final.0
python-bits: 64
OS: Linux
OS-release: 4.9.8-100.fc24.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: C
LANG: C
LOCALE: None.None

pandas: 0.19.0+531.g04e1168
pytest: 3.0.5
pip: 9.0.1
setuptools: 27.2.0
Cython: 0.25.2
numpy: 1.11.3
scipy: 0.18.1
xarray: 0.9.1
IPython: 4.2.0
sphinx: 1.5.1
patsy: 0.4.1
dateutil: 2.6.0
pytz: 2016.10
blosc: None
bottleneck: 1.2.0
tables: 3.3.0
numexpr: 2.6.2
feather: None
matplotlib: 2.0.0
openpyxl: 2.4.1
xlrd: 1.0.0
xlwt: 1.2.0
xlsxwriter: 0.9.6
lxml: 3.7.2
bs4: 4.5.3
html5lib: 0.999
sqlalchemy: 1.1.5
pymysql: None
psycopg2: None
jinja2: 2.9.4
s3fs: None
pandas_gbq: None
pandas_datareader: None

[chris-b1]

I think this may be as easy as adding a numeric_only=False to this function call. (of course with tests)

pandas/pandas/core/groupby.py

Line 1454 in 04e1168

return self._cython_transform('cummax', **kwargs)

[jreback]

This is correct and not a bug. datetimes are not numeric types. This has to be specifically enabled.
this was covered by #15054

In [29]: x = pd.DataFrame(dict(a=[1], b=pd.to_datetime(['2001'])))
    ...: >>> x.groupby('a').b.cummax(numeric_only=False)
    ...: 
    ...: 
Out[29]: 
0   2001-01-01
Name: b, dtype: datetime64[ns]

[chris-b1]

@jreback - maybe the API is set, but isn't this inconsistent?

In [23]: df = pd.DataFrame(dict(a=[1], b=pd.to_datetime(['2001'])))

In [24]: df['b'].max()
Out[24]: Timestamp('2001-01-01 00:00:00')

In [25]: df['b'].cummax()
Out[25]: 
0   2001-01-01
Name: b, dtype: datetime64[ns]

In [26]: df.groupby('a')['b'].max()
Out[26]: 
a
1   2001-01-01
Name: b, dtype: datetime64[ns]

In [27]: df.groupby('a')['b'].cummax()
Out[27]: 
DataError: No numeric types to aggregate

[jreback]

@chris-b1 hmm, ahh we are defaulting numeric_only=False for .max

yes that should be done for cummax/cummin.