QuarterEnd Offset seems to not work with datetimes on the last day

[thequackdaddy]

Hello,

I have a column of datetimes which I'm trying to group by moving them to the last day of the calendar quarter. However, I think when the datetime is on the last day of the quarter, It doesn't work.

Code Sample, a copy-pastable example if possible

In [8]: pd.Series(pd.DatetimeIndex(['2016-06-30 12:32'])).dt.date + pd.offsets.QuarterEnd()
Out[8]: 
0   2016-09-30
dtype: datetime64[ns]

Quarter Begin appears to be even more screwed up. It is moving the date to March 1, which isn't the beginning of a quarter.

pd.Series(pd.DatetimeIndex(['2016-04-01 04:02'])).dt.date - pd.offsets.QuarterBegin()
Out[9]: 
0   2016-03-01
dtype: datetime64[ns]

Expected Output

I expect this to be 2016-06-30.

output of pd.show_versions()

In [25]: pd.show_versions()

INSTALLED VERSIONS
------------------
commit: None
python: 3.5.1.final.0
python-bits: 64
OS: Windows
OS-release: 2012ServerR2
machine: AMD64
processor: Intel64 Family 6 Model 63 Stepping 2, GenuineIntel
byteorder: little
LC_ALL: None
LANG: en_US

pandas: 0.18.1
nose: 1.3.7
pip: 8.1.2
setuptools: 23.0.0
Cython: 0.24
numpy: 1.11.0
scipy: 0.17.1
statsmodels: 0.8.0.dev0+032a9ab
xarray: None
IPython: 4.2.0
sphinx: 1.3.1
patsy: 0.4.1
dateutil: 2.5.3
pytz: 2016.4
blosc: None
bottleneck: 1.0.0
tables: 3.2.2
numexpr: 2.6.0
matplotlib: 1.5.1
openpyxl: 2.3.2
xlrd: 1.0.0
xlwt: 1.1.2
xlsxwriter: 0.9.2
lxml: 3.6.0
bs4: 4.4.1
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.13
pymysql: None
psycopg2: None
jinja2: 2.8
boto: 2.40.0
pandas_datareader: None

I think a suitable workaround for QuarterEnd is to convert the datetime to a date, then subtract a day, then move to QuartrEnd. (Untested).

[sinhrks]

If I understood properly, this isn't specific for quarter. It's because we're adding the offset.

pd.Series(pd.DatetimeIndex(['2016-06-30 12:32'])).dt.date + pd.offsets.MonthEnd()
# 0   2016-07-31
# dtype: datetime64[ns]

maybe you expect .rollforward, but it doesn't support vectorization.

pd.offsets.QuarterEnd().rollforward(pd.Timestamp('2016-06-30 12:32'))
# Timestamp('2016-06-30 12:32:00')

For the second issue, you can specify startingMonth (see #8435 for discussion to change the default)

pd.Series(pd.DatetimeIndex(['2016-04-01 04:02'])).dt.date - pd.offsets.QuarterBegin(startingMonth=1)
# 0   2016-01-01
# dtype: datetime64[ns]

[jreback]

cc @chris-b1

[chris-b1]

You want an offset with n=0 - that has the semantics of not rolling the date forward if on an anchor (e.g., a QuarterEnd) and rolling forward otherwise. n=1, the default always rolls forward. See docs here

In [13]: pd.Series(pd.DatetimeIndex(['2016-06-30 12:32'])).dt.date + pd.offsets.QuarterEnd(n=0)
Out[13]:
0   2016-06-30
dtype: datetime64[ns]

[jreback]

ok closing this, but if anyone thinks docs are not clear a pr would be great.