BUG: Groupby with as_index=True causes incorrect summarization #34906
Conversation
|
Given that this is my first commit to pandas, I'm not quite sure whether what I'm doing is what is expected. I just added the example from the bugreport as a test, is that ok? Should additional things be tested? Tested differently? Please, let me know. |
jreback
changed the title
jreback
added
Groupby
Testing
pandas/tests/groupby/test_groupby.py
Outdated
| ) | ||
|
|
||
| tm.assert_series_equal( | ||
| df.groupby("b")["c"].min(), |
pandas/tests/groupby/test_groupby.py
Outdated
| tm.assert_series_equal( | ||
| df.groupby("b")["c"].min(), | ||
| df.groupby("b", as_index=False)["c"].min()["c"], | ||
| check_index_type=False, |
There was a problem hiding this comment.
if you have to set check_index_type or check_names the something is wrong with the expected here.
pandas/tests/groupby/test_groupby.py
Outdated
| date_series = pd.Series(dates) | ||
| date_series_parsed = pd.to_datetime(date_series, format="%Y-%m-%d").dt.date | ||
|
|
||
| df = pd.DataFrame( |
There was a problem hiding this comment.
pls simplify this construction, e.g. directly construct the nputs
|
I addressed the comments, could you please have a look again? |
|
I don't know why there suddenly are test failures, the particular test that failed (pandas/tests/base/test_unique.py::test_unique_bad_unicode), passes on my setup and is nowhere near the test that I added. |
|
@vivikelapoutre can you try merging master & repushing to see if that fixes it? |
Thank you, that worked indeed. |
|
Thanks @vivikelapoutre! |
…s-dev#34906) * add test * PR comments * attempt to make the code cleaner
black pandasgit diff upstream/master -u -- "*.py" | flake8 --diff