BUG: pd.concat does not drop DatetimeIndex.freq when result is monotonic with even spacing #25796
Conversation
Codecov Report
@@ Coverage Diff @@
## master #25796 +/- ##
==========================================
- Coverage 91.26% 91.26% -0.01%
==========================================
Files 173 173
Lines 52982 52983 +1
==========================================
Hits 48356 48356
- Misses 4626 4627 +1
Continue to review full report at Codecov.
|
Codecov Report
@@ Coverage Diff @@
## master #25796 +/- ##
==========================================
+ Coverage 91.48% 91.48% +<.01%
==========================================
Files 175 175
Lines 52877 52879 +2
==========================================
+ Hits 48372 48374 +2
Misses 4505 4505
Continue to review full report at Codecov.
|
pandas/core/indexes/datetimelike.py
Outdated
|
|
||
| # GH 3232: If the concat result is evenly spaced, we can retain the | ||
| # original frequency | ||
| is_diff_evenly_spaced = len(np.unique(np.diff(new_data))) == 1 |
There was a problem hiding this comment.
I think we have something that does this in tseries.frequencies, can you see / use it?
There was a problem hiding this comment.
Checked in tseries.frequencies but didn't find anything. We have is_monotonic functions but it doesn't check for even spacing.
There was a problem hiding this comment.
you can use unique_1d here (should be faster), though not a big deal. worth moving this actually to tseries.frequencies as a named function?
There was a problem hiding this comment.
this looks a bit like unique_deltas
pandas/core/indexes/datetimelike.py
Outdated
|
|
||
| # GH 3232: If the concat result is evenly spaced, we can retain the | ||
| # original frequency | ||
| is_diff_evenly_spaced = len(np.unique(np.diff(new_data))) == 1 |
There was a problem hiding this comment.
you can use unique_1d here (should be faster), though not a big deal. worth moving this actually to tseries.frequencies as a named function?
|
lgtm. merge on green. |
|
thanks @mroeschke |
3 participants
git diff upstream/master -u -- "*.py" | flake8 --diff