BUG: DataFrame.pivot drops column level names when both rows and columns are multiindexed #36655
Conversation
arw2019
force-pushed
the
GH36360
branch
4 times, most recently
from
6004334 to
fa879b7
Compare
pandas/core/groupby/generic.py
Outdated
|
|
||
| agg_axis = self._obj_with_exclusions._get_axis(1 - self.axis) | ||
| if isinstance(agg_axis, MultiIndex): | ||
| columns = Index([key.label for key in output], names=agg_axis.names) |
There was a problem hiding this comment.
does this line alone work? I thought we fixed construction for this. alt does ensure_index work here? I really do not like special casing like this anywhere except in the index module itself (and even then its not great).
There was a problem hiding this comment.
It does! I've removed the if-else clause
|
|
||
|
|
||
| def test_pivot_multiindexed_rows_and_cols(): | ||
| # GH 36360 |
There was a problem hiding this comment.
do we have appropraite tests if index OR columns are a MI and the other is not?
There was a problem hiding this comment.
We do (in this file and in test_pivot.py)
What existing tests don't cover is the case when pivot_table produces a result with multiple MutliIndexed columns - added here
jreback
added
Bug
MultiIndex
Reshaping
arw2019
force-pushed
the
GH36360
branch
2 times, most recently
from
390e9a9 to
8dd1161
Compare
pandas/core/indexes/base.py
Outdated
| @@ -404,6 +404,11 @@ def __new__( | |||
| ) | |||
| # other iterable of some kind | |||
| subarr = com.asarray_tuplesafe(data, dtype=object) | |||
|
|
|||
| if isinstance(name, list) and len(name) == 1: | |||
There was a problem hiding this comment.
we do not want to add this here, this is a very complicated routine; what are you trying to accomplish?
There was a problem hiding this comment.
I was looking for the functionality of Index._get_names() / MultiIndex._get_names()
Reverted this + using _get_names now
arw2019
force-pushed
the
GH36360
branch
3 times, most recently
from
861baa9 to
ffb4540
Compare
|
@jreback this one is ready for re-review. I now handle simple index and multiindex together with |
|
thanks @arw2019 |
…mns are multiindexed (pandas-dev#36655)
…mns are multiindexed (pandas-dev#36655)
black pandasgit diff upstream/master -u -- "*.py" | flake8 --diffFixed by checking for multi-indexed columns in
DataFrameGroupBy._wrap_aggregated_output