BUG: Grouper: Origin param has no effect

[houseofai]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd

df = pd.DataFrame([{"A": A, "datadate": datadate} 
                         for A in range(1,3) 
                         for datadate in pd.date_range(start='1/2/2022', end='2/1/2022', freq='D')])

ddg = df.groupby(["A", pd.Grouper(key="datadate", freq='W', origin='start')])

for i, dfg in ddg:
    print(dfg[["A", "datadate"]])
    print("-----------------------")


ddg = df.groupby(["A", pd.Grouper(key="datadate", freq='W', origin='1/5/2022')])

for i, dfg in ddg:
    print(dfg[["A", "datadate"]])
    print("-----------------------")

Issue Description

Whatever I set on the origin parameter of pd.Grouper, whether it is startor a date, it groups by starting on a Monday.
If I remove the origin parameter, I get the same results.

Expected Behavior

The data are grouped weekly starting by the fixed timestamp provided by the origin parameter.

Installed Versions

INSTALLED VERSIONS

commit : e8093ba
python : 3.8.13.final.0
python-bits : 64
OS : Darwin
OS-release : 21.5.0
Version : Darwin Kernel Version 21.5.0: Tue Apr 26 21:08:37 PDT 2022; root:xnu-8020.121.3~4/RELEASE_ARM64_T6000
machine : arm64
processor : arm
byteorder : little
LC_ALL : None
LANG : None
LOCALE : None.UTF-8

pandas : 1.4.3
numpy : 1.21.6
pytz : 2022.1
dateutil : 2.8.2
setuptools : 61.2.0
pip : 21.2.4
Cython : 0.29.30
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : 4.9.0
html5lib : None
pymysql : 1.0.2
psycopg2 : None
jinja2 : 3.1.1
IPython : 8.2.0
pandas_datareader: None
bs4 : 4.9.1
bottleneck : None
brotli : None
fastparquet : None
fsspec : 2022.5.0
gcsfs : None
markupsafe : 2.1.1
matplotlib : 3.5.2
numba : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 8.0.0
pyreadstat : None
pyxlsb : None
s3fs : 2022.5.0
scipy : 1.8.0
snappy : None
sqlalchemy : 1.4.32
tables : None
tabulate : 0.8.10
xarray : None
xlrd : None
xlwt : None
zstandard : None

[hamedgibago]

take

[hamedgibago]

@houseofai For the second grouper, first start day which printed should be 1 2022-01-05?

[houseofai]

@hamedgibago Yes, thanks

[DBCerigo]

Any news on this? If triage has been done could an update be given on whether it is a confirmed bug? Thanks.

[jreback]

not a bug

W is an anchored offset which starts on monday; you can use W-TUES for example as well

see https://pandas.pydata.org/docs/search.html?q=Anchored

[DBCerigo]

@jreback thanks a lot for the reply, my apologies for the time used for the correction.

[houseofai]

@jreback

df.groupby(["A", pd.Grouper(key="datadate", freq='W-TUES', origin='start')])

throws ValueError: Invalid frequency: W-TUES

[DBCerigo]

@houseofai did you try "W-TUE" as per the docs? (https://pandas.pydata.org/docs/user_guide/timeseries.html?highlight=anchored#anchored-offsets)

[houseofai]

@DBCerigo Thanks for pointing out, indeed W-TUE doesn't raise an error.

ddg = df.groupby(["A", pd.Grouper(key="datadate", freq='W-TUE')])

for i, dfg in ddg:
    print(dfg[["A", "datadate"]])
    print("-----------------------")

Gives:

   A   datadate
0  1 2022-01-02
1  1 2022-01-03
2  1 2022-01-04
-----------------------
   A   datadate
3  1 2022-01-05
4  1 2022-01-06
5  1 2022-01-07
6  1 2022-01-08
7  1 2022-01-09
8  1 2022-01-10
9  1 2022-01-11
-----------------------

Note that the first date,2022-01-05, is a Wednesday.

The only way to start the week from the first date is to find the day of the first date (here 2022-01-02 - Sunday), and to set it to the day before:

ddg = df.groupby(["A", pd.Grouper(key="datadate", freq='W-SAT')])

 A   datadate
0  1 2022-01-02
1  1 2022-01-03
2  1 2022-01-04
3  1 2022-01-05
4  1 2022-01-06
5  1 2022-01-07
6  1 2022-01-08
-----------------------
    A   datadate
7   1 2022-01-09
8   1 2022-01-10
9   1 2022-01-11
...

Could you please confirm that is the expected behavior with the frequency parameter?

And I insist but the origin parameter doesn't have any effect.

[rhshadrach]

Looking at #31809, it appears to me origin was specifically added for day and sub-day frequencies. If this is the case and it's intended to have no impact on frequencies coarser than a day, we should raise when provided (after deprecating the current behavior).