BUG: KeyError after groupby().size() when index is multiindex and involves nan

[hdou]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Note: Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

Code Sample, a copy-pastable example

# Your code here
import pandas as pd

data = [[pd.NA, 10], [pd.NA,10], [2, pd.NA]]
df = pd.DataFrame(data, columns=['A1', 'A2'])
print(df)

counts = df.groupby(['A1', 'A2'],dropna=False).size()
print(counts)

# If you uncomment the lines below, it works
# pk_file = 'tmp.df'
# counts.to_pickle(pk_file)
# counts = pd.read_pickle(pk_file)

idx = counts.index[0]
print(counts.loc[idx])  # KeyError: (2.0, nan)

Problem description

Unable to access the Series returned from groupby using apparently valid key
Line "print(counts.loc[idx]" raises KeyError.
If I uncomment the lines that store/load the data with pickle, then it works. No KeyError is raised.
[this should explain why the current behaviour is a problem and why the expected output is a better solution]

Expected Output

The Series returned from groupby is accessible with loc with valid keys

Output of pd.show_versions()

[paste the output of pd.show_versions() here leaving a blank line after the details tag]

commit : 2a7d332
python : 3.8.5.final.0
python-bits : 64
OS : Darwin
OS-release : 18.7.0
Version : Darwin Kernel Version 18.7.0: Thu Jun 18 20:50:10 PDT 2020; root:xnu-4903.278.43~1/RELEASE_X86_64
machine : x86_64
processor : i386
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.1.2
numpy : 1.19.1
pytz : 2020.1
dateutil : 2.8.1
pip : 20.1.1
setuptools : 47.1.0
Cython : 0.29.21
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : None
IPython : None
pandas_datareader: None
bs4 : None
bottleneck : None
fsspec : None
fastparquet : None
gcsfs : None
matplotlib : 3.3.1
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pytables : None
pyxlsb : None
s3fs : None
scipy : 1.5.2
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
numba : None

[phofl]

Hi,

thanks for your report. Looks like the resulting Index from the groupby operation does not treat the np.nan s within the MultiIndex as missing values. This seems to be the issue in here.

Edit: is probably a duplicate of #36060

[rhshadrach]

Thanks @phofl - agree. When you look at counts.index.codes before and after pickling, you get [[0, 1], [1, 0]] and [[0, -1], [-1, 0]] respectively. Closing as duplicate.