BUG: GroupBy return EA dtype #23318
BUG: GroupBy return EA dtype #23318
Conversation
|
Hello @5hirish! Thanks for submitting the PR.
|
|
Ran tests locally |
there is custom datatype
|
|
For failures like 1, you'll need to update
No. Int64 and int64 are different. Int64 is our integer-NA dtype. You'll need to update the Sparse tests as well. you'll need to do |
jreback
added
Groupby
ExtensionArray
|
All tests and checks have passed except for one where I got this error although all tests have passed |
|
isort error: https://travis-ci.org/pandas-dev/pandas/jobs/447147983#L2772 Make sure to merge master. LMK if you need help fixing it. |
pandas/tests/arrays/test_integer.py
Outdated
| "B": np.array([1.0, 3.0]), | ||
| "C": np.array([1, 3], dtype="int64") | ||
| }, index=pd.Index(['a', 'b'], name='A')) | ||
| tm.assert_frame_equal(result, expected) |
There was a problem hiding this comment.
fully construct the return frame as in the existing example
There was a problem hiding this comment.
Do you mean this
a)
result = df.groupby("A").op()
assert result.dtypes['C'].name == 'Int64'
b) or do you mean that I should construct the expected as:
expected = pd.DataFrame({
"B": np.array([1.0, 3.0]),
"C": np.array([1, 3], dtype="Int64")
}, index=pd.Index(['a', 'b'], name='A'))
tm.assert_frame_equal(result, expected)
Because in b) the expected constructed dataframe converts Int64 to int64.
There was a problem hiding this comment.
Because in b) the expected constructed dataframe converts Int64 to int64
The "Int64" (capital "I") is a pandas thing. You need
"C": integer_array([1, 3], dtype="Int64")
There was a problem hiding this comment.
the 2nd. we want to construct the resulting frame and compare
There was a problem hiding this comment.
Why was this marked resolved? Am I missing the expected and the tm.assert_frame_equal?
There was a problem hiding this comment.
No, you are correct, I missed that one...
pandas/tests/arrays/test_integer.py
Outdated
| "B": np.array([1.0, 3.0]), | ||
| "C": np.array([1, 3], dtype="int64") | ||
| }, index=pd.Index(['a', 'b'], name='A')) | ||
| tm.assert_frame_equal(result, expected) |
There was a problem hiding this comment.
Because in b) the expected constructed dataframe converts Int64 to int64
The "Int64" (capital "I") is a pandas thing. You need
"C": integer_array([1, 3], dtype="Int64")
|
Did |
|
@5hirish I think you needed to merge master before running isort. Pretty sure travis tests a merge commit, which may have explained the difference. I merged master and pushed a change. So you'll want to pull before making additional changes. |
Codecov Report
@@ Coverage Diff @@
## master #23318 +/- ##
==========================================
+ Coverage 92.22% 92.23% +<.01%
==========================================
Files 161 161
Lines 51191 51202 +11
==========================================
+ Hits 47210 47225 +15
+ Misses 3981 3977 -4
Continue to review full report at Codecov.
|
5hirish
force-pushed
the
bug_agg_ea_dtype
branch
from
8a9a36e
to
fc5d2f2
Compare
pandas/tests/arrays/test_integer.py
Outdated
| "B": np.array([1.0, 3.0]), | ||
| "C": np.array([1, 3], dtype="int64") | ||
| }, index=pd.Index(['a', 'b'], name='A')) | ||
| tm.assert_frame_equal(result, expected) |
There was a problem hiding this comment.
the 2nd. we want to construct the resulting frame and compare
pandas/tests/arrays/test_integer.py
Outdated
| "B": np.array([1.0, 3.0]), | ||
| "C": np.array([1, 3], dtype="int64") | ||
| }, index=pd.Index(['a', 'b'], name='A')) | ||
| tm.assert_frame_equal(result, expected) |
There was a problem hiding this comment.
Why was this marked resolved? Am I missing the expected and the tm.assert_frame_equal?
|
thanks @5hirish |
git diff upstream/master -u -- "*.py" | flake8 --diff