Add unstack() support for non-multiindexed dataframes
#7054
Conversation
isVoid
added
Python
Codecov Report
@@ Coverage Diff @@
## branch-0.18 #7054 +/- ##
===============================================
+ Coverage 82.09% 82.12% +0.02%
===============================================
Files 97 97
Lines 16477 16487 +10
===============================================
+ Hits 13527 13540 +13
+ Misses 2950 2947 -3
Continue to review full report at Codecov.
|
python/cudf/cudf/core/reshape.py
Outdated
|
|
||
| Unstacking single level index dataframe: | ||
|
|
||
| >>> df.unstack(['b', 'd']).unstack() |
There was a problem hiding this comment.
I think this example is a little opaque - it's sometimes difficult to visualize exactly what the result of unstack should be for even a single level, and here I find it a little hard to connect to dots through the chained operation. I'd recommend an example that starts with a dataframe with a single index and shows the result of unstacking that dataframe into a series instead.
python/cudf/cudf/core/reshape.py
Outdated
| "is not supported" | ||
| ) | ||
| if isinstance(df, cudf.DataFrame): | ||
| res = df.T.stack(dropna=False) |
There was a problem hiding this comment.
Does this pass the typecasting behavior off to transpose? Should we check the dtypes and possibly error here?
There was a problem hiding this comment.
It seems like both transpose.pyx and libcudf::transpose.cu checks whether all columns have the same datatype. A clear exception gets raised if the columns are of different types. Should we check again here?
There was a problem hiding this comment.
I would support checking here - imagining what happens here from the user perspective, if I get an error trying to unstack a cuDF dataframe, I might wonder why the transpose code is unhappy.
In general, I think we try and avoid letting libcudf itself serve an error to the user and favor a more surface level python error, usually when I've managed to actually manifest a libcudf error from the python API it means something is very wrong.
python/cudf/cudf/core/reshape.py
Outdated
| ) | ||
| if isinstance(df, cudf.DataFrame): | ||
| dtype = df._columns[0].dtype | ||
| if any( |
There was a problem hiding this comment.
It won't matter much in most cases but I think it's slightly more efficient to use a single loop that raises if df._columns[i].dtype != dtype as opposed to creating a list with all the booleans and then doing an any reduction.
There was a problem hiding this comment.
Raised an issue since this comes up in a bunch of places #7067
harrism
changed the title
unstack() support for non-multiindexed dataframesunstack() support for non-multiindexed dataframes
|
rerun tests |
isVoid
added
6 - Okay to Auto-Merge
and removed
3 - Ready for Review
Closes #6694
When
unstack()receives a dataframe with "single" index, returns a series to match pandas behavior.