BUG: Avoid casting to double type unnecessarily when setting values i… #23462
Conversation
|
Hello @icexelloss! Thanks for submitting the PR.
|
icexelloss
force-pushed
the
timedelta-casting-bug
branch
2 times, most recently
from
b43b803 to
ca6bbd3
Compare
Codecov Report
@@ Coverage Diff @@
## master #23462 +/- ##
=======================================
Coverage 92.23% 92.23%
=======================================
Files 161 161
Lines 51197 51197
=======================================
Hits 47220 47220
Misses 3977 3977
Continue to review full report at Codecov.
|
jreback
added
Bug
Indexing
jreback
left a comment
jreback
left a comment
There was a problem hiding this comment.
small comments. pls add a whatsnew note in bug fixes / indexing. ping on green.
| tm.assert_series_equal(result, expected) | ||
|
|
||
| def test_set_dataframe_column_by_index(self): | ||
|
|
There was a problem hiding this comment.
can you add this PR number here as a comment
| expected = s.iloc[expected_slice] | ||
| tm.assert_series_equal(result, expected) | ||
|
|
||
| def test_set_dataframe_column_by_index(self): |
There was a problem hiding this comment.
can rename to
test_roundtrip_thru_setitem
|
|
||
| df = pd.DataFrame({'dt': pd.Series([dt1, dt2])}) | ||
| s = pd.Series([dt1]) | ||
| value_before = df['dt'].iloc[1].value |
| s = pd.Series([dt1]) | ||
| value_before = df['dt'].iloc[1].value | ||
| df.loc[[True, False]] = s | ||
| value_after = df['dt'].iloc[1].value |
| value_before = df['dt'].iloc[1].value | ||
| df.loc[[True, False]] = s | ||
| value_after = df['dt'].iloc[1].value | ||
|
|
There was a problem hiding this comment.
can you also test
tm.assert_frame_equal(df, df_copy)
(and copy df on line 89)
icexelloss
force-pushed
the
timedelta-casting-bug
branch
from
ca6bbd3 to
62ab354
Compare
gfyoung
added
Numeric Operations
|
lgtm. ping on green. |
|
thanks @icexelloss |
|
Thanks @jreback ! |
4 participants
…n time delta column
git diff upstream/master -u -- "*.py" | flake8 --diffWhen setting a timedelta column in a dataframe using
pd.Seriesor array as index, there is some rounding error because the underlying TimeDeltaBlock is casted to double and back to int. This patch fixes the issue.