BUG: dtype conversion fails on loc after pivot

[ljmc-github]

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

pd.__version__  # 2.0.0

df1 = pd.DataFrame(
    {
        "n": [11, 13, 17],
        "answers": ["A", "A", "B"],
    }
)

df1 = df1.pivot(columns="answers", values="n")

df1.loc[:, ["A", "B"]] = df1.loc[:, ["A", "B"]].fillna(0).astype("int")

# df1
# answers     A     B
# 0        11.0   0.0
# 1        13.0   0.0
# 2         0.0  17.0

df2 = pd.DataFrame({
    "A": [11.0, 13.0, pd.NA],
    "B": [pd.NA, pd.NA, 17.0]
})

df2.loc[:, ["A", "B"]] = df2.loc[:, ["A", "B"]].fillna(0).astype("int")

# df2
#     A   B
# 0  11   0
# 1  13   0
# 2   0  17

Issue Description

The expression df1.loc[:, ["A", "B"]].fillna(0).astype("int") gives the expected data with NAs filled with 0 and the dtype changed to int.

But assigning that back to df1.loc[:, ["A", "B"]] does not keep the dtype change.

I have also verified that copying df2's column into df1's columns before the fillna and astype does not change the outcome.

This is not reproducible in a dataframe created from the pivoted data (df2).

NB.
This MRE is a much simplified version of the indexing/assignment I use in the code in which I found the issue, which uses pd.IndexSlice and hierarchical columns.

I am aware df1[["A", "B"]] = df1[["A", "B"]].fillna(0).astype("int") works but using .loc is necessary with pd.IndexSlice.

I am currently using a workaround similar to the above by creating an exhaustive pd.MultiIndex and using it in place of ["A", "B"] rather than using a .loc + pd.IndexSlice, happy to work on a hierarchical MRE if that's useful or necessary.

Expected Behavior

The pivoted df1 should behave like df2 and astype not be lost upon assignment.

Installed Versions

INSTALLED VERSIONS

commit : 478d340
python : 3.9.16.final.0
python-bits : 64
OS : Darwin
OS-release : 22.3.0
Version : Darwin Kernel Version 22.3.0: Mon Jan 30 20:42:11 PST 2023; root:xnu-8792.81.3~2
/RELEASE_X86_64
machine : x86_64
processor : i386
byteorder : little
LC_ALL : en_GB.UTF-8
LANG : en_GB.UTF-8
LOCALE : en_GB.UTF-8

pandas : 2.0.0
numpy : 1.24.2
pytz : 2023.3
dateutil : 2.8.2
setuptools : 58.1.0
pip : 23.0.1
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : None
IPython : 8.12.0
pandas_datareader: None
bs4 : None
bottleneck : None
brotli : None
fastparquet : None
fsspec : None
gcsfs : None
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : 3.1.2
pandas_gbq : None
pyarrow : None
pyreadstat : None
pyxlsb : None
s3fs : None
scipy : None
snappy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
zstandard : None
tzdata : 2023.3
qtpy : None
pyqt5 : None

[ryanyao2002]

take

[phofl]

Hey, thanks for your report. This is a duplicate of #52593