Pandas BUG: Losing information when handling of NoneTypes in pd.json_normalize

Pandas version checks

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

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

• [X] I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

def flatten_json(data: dict):
    sr = pd.Series(data)
    df = pd.DataFrame()
    df["jsonData"] = sr
    flat = pd.json_normalize(df['jsonData'])
    return df.join(flat)

expectedData = {
    0: {'Field1': 'SomeValue'},
    1: {'Field1': None},
    2: {},
}
expectedDf = flatten_json(expectedData)
display(expectedDf)

buggedData = {
    0: {},
    1: {'Field1': None},
    2: {},
}
buggedDf = flatten_json(buggedData)
display(buggedDf)

Issue Description

When using pd.json_normalize I expected to keep the distinction between None and NaN cells (or at the very least for the behaviour to be the same regardless of the input.

There are 2 scenarios in the example I gave: Scenario 1 - At least one value for a specific key (resultant column in the df) in the json is not None or NaN. In this case the cells containing None are preserved in the final df

Scenario 2 - All the values for a key (resultant column in the df) in the json payload are None or NaN. In this case the cells containing None are not preserved in the final df, rather None is converted to NaN

The handling of None should at the very least be consistent in both scenarios. In my opinion it should retain as much information as possible, so the output of scenario 1 makes the most sense to me.

Expected Behavior

I expect to still see a NoneType value even when all the cells are either None or NaN

Installed Versions

INSTALLED VERSIONS ------------------ commit : d9cdd2ee5a58015ef6f4d15c7226110c9aab8140 python : 3.11.9.final.0 python-bits : 64 OS : Windows OS-release : 10 Version : 10.0.22631 machine : AMD64 processor : Intel64 Family 6 Model 140 Stepping 1, GenuineIntel byteorder : little LC_ALL : None LANG : None LOCALE : English_Australia.1252 pandas : 2.2.2 numpy : 1.26.4 pytz : 2024.1 dateutil : 2.9.0 setuptools : 68.2.2 pip : 23.3.1 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 : 8.22.2 pandas_datareader : None adbc-driver-postgresql: None adbc-driver-sqlite : None bs4 : None bottleneck : None dataframe-api-compat : None fastparquet : None fsspec : None gcsfs : None matplotlib : None numba : None numexpr : None odfpy : None openpyxl : None pandas_gbq : None pyarrow : None pyreadstat : None python-calamine : None pyxlsb : None s3fs : None scipy : None sqlalchemy : None tables : None tabulate : None xarray : None xlrd : None zstandard : None tzdata : 2024.1 qtpy : None pyqt5 : None

Comment From: nguyen-tien-tung

take

Comment From: a-theron

After reviewing the documentation, I think this might not be a bug. I think it is a result of how Pandas handles none types. In Panda's none type documentation it states

pandas uses different sentinel values to represent a missing (also referred to as NA) depending on the data type.

Which means that the different none types (None, NaN, NaT) are used depending on the dtype of the column. So when a column only has missing values (as shown in the "bug"), Pandas has no reference dtypes so all missing values are represented by the default (NaN) none type.

So correct me if I am wrong, but I believe this might actually part of the Panda's feature set, rather than a bug.