BUG: Styling of a `DataFrame` with boolean column labels leaves out `False` and fails when adding non-numerical column · Issue #47838 · pandas-dev/pandas (original) (raw)

Pandas version checks

I have checked that this issue has not already been reported.
I have confirmed this bug exists on the latest version of pandas.
I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

%%

import pandas as pd from IPython import display

df = pd.DataFrame( [ [1, 2], [3, 4] ], columns=[False, True], index=[1, 2], )

column `False` not styled (requires running in Jupyter)

display.HTML(df.style.background_gradient().to_html())

%%

df["C"] = ["a", "b"]

fails with IndexError

display.HTML(df.style.background_gradient().to_html())

Issue Description

Before adding C, the styled dataframe renders like this:

After adding C, I get IndexError: Boolean index has wrong length: 2 instead of 3.

Traceback

Traceback (most recent call last):
  File "minimal.py", line 18, in <module>
    display.HTML(df.style.background_gradient().to_html())
  File "/home/pandas/pandas/io/formats/style.py", line 1372, in to_html
    html = obj._render_html(
  File "/home/pandas/pandas/io/formats/style_render.py", line 202, in _render_html
    d = self._render(sparse_index, sparse_columns, max_rows, max_cols, "&nbsp;")
  File "/home/pandas/pandas/io/formats/style_render.py", line 166, in _render
    self._compute()
  File "/home/pandas/pandas/io/formats/style_render.py", line 254, in _compute
    r = func(self)(*args, **kwargs)
  File "/home/pandas/pandas/io/formats/style.py", line 1711, in _apply
    data = self.data.loc[subset]
  File "/home/pandas/pandas/core/indexing.py", line 1065, in __getitem__
    return self._getitem_tuple(key)
  File "/home/pandas/pandas/core/indexing.py", line 1254, in _getitem_tuple
    return self._getitem_tuple_same_dim(tup)
  File "/home/pandas/pandas/core/indexing.py", line 922, in _getitem_tuple_same_dim
    retval = getattr(retval, self.name)._getitem_axis(key, axis=i)
  File "/home/pandas/pandas/core/indexing.py", line 1287, in _getitem_axis
    return self._getbool_axis(key, axis=axis)
  File "/home/pandas/pandas/core/indexing.py", line 1089, in _getbool_axis
    key = check_bool_indexer(labels, key)
  File "/home/pandas/pandas/core/indexing.py", line 2561, in check_bool_indexer
    return check_array_indexer(index, result)
  File "/home/pandas/pandas/core/indexers/utils.py", line 552, in check_array_indexer
    raise IndexError(
IndexError: Boolean index has wrong length: 2 instead of 3

The same problem occurs when using style.bar instead of style.background_gradient.

Expected Behavior

First table:

After adding column "C":

I think the problem is that subset is by default taken to be the column names of the numerical columns of df

if subset is None and gmap is None:
subset = self.data.select_dtypes(include=np.number).columns

Later, df is indexed with loc, which treats subset as a boolean mask:

subset = non_reducing_slice(subset)
data = self.data.loc[subset]

A solution could be to make subset by default a boolean mask:

        subset = (
            self.data.columns.isin(
                self.data.select_dtypes(include=np.number)
            )
        )

I tried this replacement and used it to produce the screenshots for the expected behavior above.

I suppose having boolean column labels is exotic and probably not a good idea generally. When indexing, for example df[[False]], pandas interprets this as a boolean mask. Yet, I encountered this problem when trying to visualize pd.crosstab(a, b), where b has boolean values.

If there is interest in fixing this, I'd be happy to work on a PR. Perhaps the very least, a warning should be shown when styling with boolean column labels.

Installed Versions

/opt/conda/lib/python3.8/site-packages/_distutils_hack/init.py:33: UserWarning: Setuptools is replacing distutils.
warnings.warn("Setuptools is replacing distutils.")

INSTALLED VERSIONS

commit : a62897a
python : 3.8.13.final.0
python-bits : 64
OS : Linux
OS-release : 5.10.104-linuxkit
Version : #1 SMP Thu Mar 17 17:08:06 UTC 2022
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : C.UTF-8
LANG : C.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.5.0.dev0+1189.ga62897ae5f
numpy : 1.22.4
pytz : 2022.1
dateutil : 2.8.2
setuptools : 63.2.0
pip : 22.2
Cython : 0.29.30
pytest : 7.1.2
hypothesis : 6.47.1
sphinx : 4.5.0
blosc : None
feather : None
xlsxwriter : 3.0.3
lxml.etree : 4.9.1
html5lib : 1.1
pymysql : 1.0.2
psycopg2 : 2.9.3
jinja2 : 3.1.2
IPython : 8.4.0
pandas_datareader: 0.10.0
bs4 : 4.11.1
bottleneck : 1.3.5
brotli :
fastparquet : 0.8.1
fsspec : 2022.5.0
gcsfs : 2022.5.0
matplotlib : 3.5.2
numba : 0.55.2
numexpr : 2.8.3
odfpy : None
openpyxl : 3.0.9
pandas_gbq : 0.17.6
pyarrow : 8.0.0
pyreadstat : 1.1.9
pyxlsb : 1.0.9
s3fs : 0.6.0
scipy : 1.8.1
snappy :
sqlalchemy : 1.4.39
tables : 3.7.0
tabulate : 0.8.10
xarray : 2022.6.0
xlrd : 2.0.1
xlwt : 1.3.0
zstandard : 0.18.0

BUG: Styling of a DataFrame with boolean column labels leaves out False and fails when adding non-numerical column · Issue #47838 · pandas-dev/pandas (original) (raw)