TST: added test for GH28597 #33588
TST: added test for GH28597 #33588
Conversation
Added test to ensure that categories stay ordered when grouping with missing values.
pandas/tests/groupby/test_groupby.py
Outdated
| @@ -2012,3 +2012,48 @@ def test_groups_repr_truncates(max_seq_items, expected): | |||
|
|
|||
| result = df.groupby(np.array(df.a)).groups.__repr__() | |||
| assert result == expected | |||
|
|
|||
|
|
|||
| def test_sorted_missing_category_values(): | |||
There was a problem hiding this comment.
can you move this test to pandas/tests/groupby/test_categorical.py
There was a problem hiding this comment.
No problem, will do
pandas/tests/groupby/test_groupby.py
Outdated
| dtype="category", | ||
| ) | ||
|
|
||
| result = df.groupby(["bar", "foo"]).size().unstack() |
There was a problem hiding this comment.
I think the change in behaviour that needs to be tested is the size method and the unstack call is not relevant in the issue OP.
There was a problem hiding this comment.
There are two separate issues that have been fixed, and I think that this tests them both. One is that missing categorical values in groupby would cause the categories to become unsorted #28597 . That is the one that I have mentioned in the commit message.
The other is that previously, empty groups would be missing from the output even when grouping with categories #23865, which I think this test also catches as a side effect.
There was a problem hiding this comment.
IIUC it's the size call that is the root cause of the issues and this issue was resolved in #29690
which I think this test also catches as a side effect
maybe one of the issues should be closed as a duplicate?
simonjayhawkins
added
the
Testing
Added test to ensure that categories stay ordered when grouping
with missing values.
black pandasgit diff upstream/master -u -- "*.py" | flake8 --diff