API: offsets.Day is always calendar-day #61985
Conversation
|
is stumping me. Who is our go-to person for debugging these? |
|
Still seeing Tried building the docs locally and jinja2 is complaining |
|
Looks like there was even more whitespace I messed up. It's happy now. |
| @@ -2763,7 +2768,10 @@ def _adjust_dates_anchored( | |||
| if offset is not None: | |||
| offset = offset.as_unit(unit) | |||
|
|
|||
| freq_value = Timedelta(freq).as_unit(unit)._value | |||
| if isinstance(freq, Day): | |||
There was a problem hiding this comment.
In #41943 I was hoping that if D means calendar day, we wouldn't need to force to be a fixed frequency in resample
There was a problem hiding this comment.
yah i was hoping for this too but wasn't able to get it working. will give it another try after addressing other comments
There was a problem hiding this comment.
just double-checked. im pretty sure the place to change the treatment would be in _get_timestamp_range_edges a bit up the call stack. Changing that to have Day cases no longer go down the Tick route breaks test_resample_origin_with_day_freq_on_dst and test_resample_tz_localized2. it wasn't obvious that the "new" behavior was correct in those tests with that change. can you confirm?
There was a problem hiding this comment.
im pretty sure the place to change the treatment would be in _get_timestamp_range_edges
That makes sense to me, as I would expect us not to have to special case Day anymore
test_resample_origin_with_day_freq_on_dst and test_resample_tz_localized2.
Mind showing an example of the new behavior based on those tests? I would expect breakage for these tests since D changing to calendar day impacts the results around DST.
There was a problem hiding this comment.
Based on test_resample_tz_localized2:
idx = date_range("2001-09-20 15:59", "2001-09-20 16:00", freq="min", tz="Australia/Sydney")
s = Series([1, 2], index=idx)
result = s.resample("D", closed="right", label="right").mean()
ex_index = date_range("2001-09-21", periods=1, freq="D", tz="Australia/Sydney")
expected = Series([1.5], index=ex_index)
tm.assert_series_equal(result, expected) # Fails with actual result being...
ex_index = pd.date_range("2001-09-20", "2001-09-21", tz="Australia/Sydney", freq="D")
expected = Series([np.nan, 1.5], index=ex_index)
doc/source/whatsnew/vX.X.X.rstfile if fixing a bug or adding a new feature.Alternative to #55502 discussed at last week's dev meeting. This allows TimedeltaIndex.freq to be a
Dayeven though it is not a Tick.