BUG: Coercing bool types to int in qcut #28802
Conversation
pandas/core/reshape/tile.py
Outdated
| @@ -444,7 +444,7 @@ def _coerce_to_type(x): | |||
| if dtype is not None: | |||
| # GH 19768: force NaT to NaN during integer conversion | |||
| if is_bool_dtype(x): | |||
There was a problem hiding this comment.
@jreback Not entirely sure where this should go - Adding x.astype(int) under elif is_bool_dtype(x) higher up throws an error in tests with the existing np.where(x.notna(), x.view(np.int64), np.nan) statement if x is ndarray - it passes if x is Series though.
AttributeError: 'numpy.ndarray' object has no attribute 'isnan'
It passes if adding is_bool_dtype(x) with new np.where condition using ~np.isnan(x) like i've done here to account for x being an ndarray
There was a problem hiding this comment.
so don't use np.isnan, change this to
x = np.where(notna(x), x.astype(np.int64, copy=False), np.nan)
There was a problem hiding this comment.
I've gone with doing the integer conversion in the elif block higher up and leaving dtype as None as @jschendel suggested, rather than making any changes here.
ryankarlos
changed the title
jschendel
added
Bug
Dtype Conversions
pandas/core/reshape/tile.py
Outdated
| @@ -444,7 +444,7 @@ def _coerce_to_type(x): | |||
| if dtype is not None: | |||
| # GH 19768: force NaT to NaN during integer conversion | |||
| if is_bool_dtype(x): | |||
There was a problem hiding this comment.
so don't use np.isnan, change this to
x = np.where(notna(x), x.astype(np.int64, copy=False), np.nan)
doc/source/whatsnew/v0.25.1.rst
Outdated
| @@ -110,6 +110,7 @@ Other | |||
|
|
|||
| - Bug in :meth:`Series.replace` and :meth:`DataFrame.replace` when replacing timezone-aware timestamps using a dict-like replacer (:issue:`27720`) | |||
| - Bug in :meth:`Series.rename` when using a custom type indexer. Now any value that isn't callable or dict-like is treated as a scalar. (:issue:`27814`) | |||
| - :func:`qcut` and `cut` now handle boolean input (:issue:`20303`) | |||
There was a problem hiding this comment.
can you put :func:`cut` as well
…ercing-bool-types-to-int-in-qcut
|
thanks @ryankarlos |
pytest pandas/tests/reshape/test_qcut.py pandas/tests/reshape/test_cut.py -vblack pandasgit diff upstream/master --name-only -- "*.py" | xargs flake8