'closed' parameter is not working for Rolling with duplicate Timestamp

[jiehuan]

The 'closed' parameter is not really working when the date is duplicated.

Here is an example.

import pandas as pd

df = pd.DataFrame({'Date': ['2016-06-30', '2016-06-30', '2016-08-09'],
                   'amount': [100, 200, 10]
                  })

df['Date'] = pd.to_datetime(df['Date'])

df

	Data	amount
0	2016-06-30	100
1	2016-06-30	200
2	2016-08-09	10

df.rolling(window='90d', on='Date', closed='neither').sum()

Output

	Data	amount
0	2016-06-30	NaN
1	2016-06-30	100
2	2016-08-09	300

Expected output:

	Data	amount
0	2016-06-30	NaN
1	2016-06-30	NaN
2	2016-08-09	300

Output of pd.show_versions()

show_version

INSTALLED VERSIONS

commit: None
python: 3.7.0.final.0
python-bits: 64
OS: Darwin
OS-release: 17.7.0
machine: x86_64
processor: i386
byteorder: little
LC_ALL: en_US.UTF-8
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8

pandas: 0.23.4
pytest: None
pip: 18.1
setuptools: 40.4.3
Cython: None
numpy: 1.15.2
scipy: 1.1.0
pyarrow: None
xarray: None
IPython: 7.0.1
sphinx: None
patsy: None
dateutil: 2.7.3
pytz: 2018.5
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: 3.0.0
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[TomAugspurger]

Could you update your example to be reproducible (steps to crate the dataframe, execute the problematic method)? http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports

And include the output of show_versions in a details tag.

[jiehuan]

updated. Thanks!

[TomAugspurger]

Thanks. It's not immediately clear to me what's going on here. If you're able to do any additional debugging, it'd be appreciated. It's probably an issue somewhere in pandas/core/window.py, or (more likely) in pandas/_libs/window.pyx.

[mroeschke]

Closing as a duplicate of #20712