TST: added test for groupby when calling first, last, nth (GH29645) #43358
Conversation
pandas/tests/groupby/test_groupby.py
Outdated
|
|
||
|
|
||
| @pytest.mark.parametrize("method", ["first", "last", "nth"]) | ||
| def test_groupby_last_first_nth_with_none(method): |
There was a problem hiding this comment.
Please define expected without applying a method and then parametrize over None and nan
There was a problem hiding this comment.
I don't quite understand what you mean: the issue states that None and np.nan should lead to the same result when I groupby and apply a method. If I don't apply a method, then what's the point of the test? My reasoning was: I have two Series with None and np.nan and I have to test that grouping with a method gives the same output.
There was a problem hiding this comment.
We would want to test the expected output explicitly. If both yield the same, the result may be incorrect nevertheless
There was a problem hiding this comment.
Ok, I think I get what you mean
pandas/tests/groupby/test_groupby.py
Outdated
| expected = Series([True]) | ||
| data = ( | ||
| Series([nans, "x", nans, "y", nans], index=[0, 0, 0, 0, 0]) | ||
| .isna() |
There was a problem hiding this comment.
I don't think isna is necessary.
Does parametrization over all missing values work? pd.NA, NaT etc
There was a problem hiding this comment.
If you mean something like @pytest.mark.parametrize("nans", [None, np.nan, pd.NA, pd.NaT]) without isna then no, all tests fail. They all pass if isna is there
There was a problem hiding this comment.
If you check the same as in #29645 (comment)? We do not want to check for True False rather than for "y".
We have a missing value fixture, but yes I meant like this
There was a problem hiding this comment.
That's reasonable, I misunderstood
jreback
added
Groupby
Testing
pandas/tests/groupby/test_groupby.py
Outdated
| @@ -2419,3 +2419,20 @@ def test_rolling_wrong_param_min_period(): | |||
| result_error_msg = r"__init__\(\) got an unexpected keyword argument 'min_period'" | |||
| with pytest.raises(TypeError, match=result_error_msg): | |||
| test_df.groupby("name")["val"].rolling(window=2, min_period=1).sum() | |||
|
|
|||
|
|
|||
| @pytest.mark.parametrize("method", ["first", "last", "nth"]) | |||
There was a problem hiding this comment.
can you move to test_nth.py
|
thanks @lorenzophys |
Provided test for
groupbywhen calllingfirst,last,nthon aSerieswithNoneornp.nan.Tests and linter pass locally.