BUG: GH11616 fixes timezone selection error #11672
Conversation
| @@ -1321,6 +1321,8 @@ def _possibly_downcast_to_dtype(result, dtype): | |||
| try: | |||
| result = result.astype(dtype) | |||
| except: | |||
| # convert to datetime and change timezone | |||
There was a problem hiding this comment.
this is too general. it can be a datetimelike (e.g. either M or m), and may or may not have a timezone. I think more tests are in order.
|
pls add a note in 0.18.0 bug fixes as well. |
|
pls add a test for #11682 which I believe this should fix as well |
varunkumar-dev
force-pushed
the
fix_groupby_timezone
branch
from
2028eb3 to
cd40984
Compare
|
These errors are occurring for different reasons. First error #11616 is in _possibly_downcast_to_dtype. It is not able to retrieve correct date from int64 object. hence throwing an exception that I have fixed by checking for timezone info and then converting to correct timezone. |
| @@ -1321,6 +1321,9 @@ def _possibly_downcast_to_dtype(result, dtype): | |||
| try: | |||
| result = result.astype(dtype) | |||
| except: | |||
| if dtype.tz: | |||
| # convert to datetime and change timezone | |||
| result = pd.to_datetime(result).tz_localize(dtype.tz) | |||
| pass | |||
|
looks good. pls add 2 entries in bug fixes in whatsnew for 0.18.0 |
varunkumar-dev
force-pushed
the
fix_groupby_timezone
branch
from
cd40984 to
fadb72d
Compare
|
Done. Please review. |
| @@ -105,3 +105,6 @@ Performance Improvements | |||
|
|
|||
| Bug Fixes | |||
| ~~~~~~~~~ | |||
|
|
|||
| - Bug in ``com._possibly_downcast_to_dtype`` not able to convert int64 object to correct datetime (:issue:`11616`) | |||
There was a problem hiding this comment.
no, use plain language from a user perspective; e.g. don't reference internal functions
Fixes GH11682 timzone info lost
varunkumar-dev
force-pushed
the
fix_groupby_timezone
branch
from
fadb72d to
f6daf50
Compare
|
merged via e838266 thanks! |
closes #11616
closes #11682