BUG: replace with np.nan unexpectedly converts pd.Timestamp to pd.NaT

[tanjt107]

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
import numpy as np

data = {
    "date": [
        pd.Timestamp("2025-01-01"),
        pd.Timestamp("2025-01-02"),
        pd.Timestamp("2025-01-03"),
    ],
}

df = pd.DataFrame(data)

df.replace([pd.Timestamp("2025-01-01"), pd.Timestamp("2025-01-02")], np.nan)

Issue Description

When using DataFrame.replace() to replace specific pd.Timestamp values with np.nan, the resulting values become pd.NaT instead of np.nan. This behavior differs from pandas 1.1.5, where the replaced values were np.nan as expected.

Output

        date
0        NaT
1        NaT
2 2025-01-03

Expected Behavior

        date
0        NaN
1        NaN
2 2025-01-03

Installed Versions

INSTALLED VERSIONS

commit : 0691c5c
python : 3.11.5
python-bits : 64
OS : Darwin
OS-release : 24.3.0
Version : Darwin Kernel Version 24.3.0: Thu Jan 2 20:24:16 PST 2025; root:xnu-11215.81.4~3/RELEASE_ARM64_T6000
machine : arm64
processor : arm
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 2.2.3
numpy : 2.0.0
pytz : 2024.1
dateutil : 2.9.0.post0
pip : 23.2.1
Cython : None
sphinx : None
IPython : None
adbc-driver-postgresql: None
adbc-driver-sqlite : None
bs4 : None
blosc : None
bottleneck : None
dataframe-api-compat : None
fastparquet : None
fsspec : None
html5lib : None
hypothesis : None
gcsfs : None
jinja2 : 3.1.4
lxml.etree : None
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
psycopg2 : None
pymysql : None
pyarrow : None
pyreadstat : None
pytest : None
python-calamine : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlsxwriter : None
zstandard : None
tzdata : 2025.2
qtpy : None
pyqt5 : None

[rhshadrach]

Thanks for the report!

This behavior differs from pandas 1.1.5, where the replaced values were np.nan as expected.

If we were to store np.nan, I think that the column dtype would have to be object. Assuming that is the case, this would be less performant and not suppor the .dt namespace that provides much of the functionality associated with timestamps.

If you do really want this behavior, then you have to opt into object dtype explicity:

df.astype(object).replace([pd.Timestamp("2025-01-01"), pd.Timestamp("2025-01-02")], np.nan)

That will give you np.nan as expected.

It seems to me the current behavior is preferred. Closing.