Rolling with offset and axis=1

[carlr67]

Code Sample

Here I have a DataFrame with a DatetimeIndex for the columns:

cols = pd.date_range('2019-8-01', '2019-08-07', freq='D')
data = np.ones((5, len(cols)))
df = pd.DataFrame(data, columns=cols)
df.rolling('1d', axis=1).sum()

The above gives 'ValueError: window must be an integer'

Here I have a DataFrame with a DatetimeIndex for the rows:

df2 = df.transpose()
df2.rolling('2d', axis=0).sum()

However this works totally fine.

Problem description

I believe there may be a bug: If I have a DatetimeIndex in the columns, specifying an offset for df.rolling does not work but gives a value error (that the window must be an integer). It works however as expected if I transpose the dataframe.

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.7.4.final.0
python-bits : 64
OS : Windows
OS-release : 10
machine : AMD64
processor : Intel64 Family 6 Model 158 Stepping 10, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : None.None

pandas : 0.25.1
numpy : 1.16.4
pytz : 2019.2
dateutil : 2.8.0
pip : 19.2.2
setuptools : 41.0.1
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.10.1
IPython : 7.7.0
pandas_datareader: None
bs4 : None
bottleneck : None
fastparquet : None
gcsfs : None
lxml.etree : None
matplotlib : 3.1.0
numexpr : None
odfpy : None
openpyxl : 2.6.2
pandas_gbq : None
pyarrow : None
pytables : None
s3fs : None
scipy : 1.3.1
sqlalchemy : None
tables : None
xarray : None
xlrd : 1.2.0
xlwt : None
xlsxwriter : None

[WillAyd]

Our support for rolling across axis=1 is less tested and feature complete than axis=0 so not terribly surprising. If you'd like to investigate and submit a PR would certainly be welcome

[MarcoGorelli]

In your example with axis=0, then in Rolling.validate from pandas/core/windows/rolling.py, self.is_datetimelike returns True, while for axis=1 it returns False. If you're not working on this, I'd be interested in giving it a go

[carlr67]

Thanks for finding that out. Sure, feel free to give it a go!