BUG: pandas.DataFrame.apply generates unexpected behavior when the np.nan is the first value in a column

[jschubnell]

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_1 = {"a": [1, 2, 3, 4],
          "b": [np.nan, -0.475, 13.795, -574]}

df_1 = pd.DataFrame(data=data_1)

result_1 = df_1.apply(min, axis=0)

print(result_1)

data_2 = {"a": [2, 1, 3, 4],
          "b": [-0.475, np.nan, 13.795, -574]}

df_2 = pd.DataFrame(data=data_2)

result_2 = df_2.apply(min, axis=0)

print(result_2)

Issue Description

Hello,

I am using the Python built-in function min() to find the smallest item in each column of my dataframe. The issue I am facing is that whenever the first value in the column b is np.nan the result is that np.nan is the smallest value when it should be -574.0. The unexpected behavior happens when I change the order of the rows, like df_2 and it returns the correct -574.0 value.

Expected Behavior

I would expect that the output for the min() value in column b is -574.0 regardless of whether the first value in the column is np.nan.
If I replace the min() for np.nanmin() I get the expected value in both cases

import pandas as pd
import numpy as np

data_1 = {"a": [1, 2, 3, 4],
         "b": [np.nan, -0.475, 13.795, -574]}

df_1 = pd.DataFrame(data=data_1)

result_1 = df_1.apply(np.nanmin, axis=0)

print(result_1)

Output:

a     1
b   -574.0
dtype: float64

data_2 = {"a": [2, 1, 3, 4],
          "b": [-0.475, np.nan, 13.795, -574]}

df_2 = pd.DataFrame(data=data_2)

result_2 = df_2.apply(np.nanmin, axis=0)

print(result_2)

Output:

a    1
b   -574.0
dtype: float64

Installed Versions

INSTALLED VERSIONS

commit : 0f43794
python : 3.10.6.final.0
python-bits : 64
OS : Linux
OS-release : 5.19.0-46-generic
Version : #47~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Jun 21 15:35:31 UTC 2
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : pt_BR.UTF-8

pandas : 2.0.3
numpy : 1.25.1
pytz : 2023.3
dateutil : 2.8.2
setuptools : 59.6.0
pip : 22.0.2
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 : None
pandas_datareader: None
bs4 : None
bottleneck : None
brotli : None
fastparquet : None
fsspec : None
gcsfs : None
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : None
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

[harikirank]

There seems to be a problem with the inbuilt min function itself. I have seen that, internally it is taking the column values and calling the inbuilt python min function. I would like to work on this issue if anyone is willing to guide me.
Thanks

[jschubnell]

Thanks for the feedback @harikirank! I would like to work on this issue too if possible.

[rhshadrach]

Related: #53425

I have seen that, internally it is taking the column values and calling the inbuilt python min function.

This is correct - apply takes a callable and uses it. This is what Python's min function does:

print(min([np.nan, -574.0]))
# nan
print(min([-574.0, np.nan]))
# -574.0

I think this is what should be expected by passing min to apply - you get exactly the behavior of Python's min builtin.

On main, you get -574.0 by passing "min" (the string alias). I think you should be passing this instead.

[jschubnell]

Thank you for the response @rhshadrach! I will close this issue as indeed the apparent inconsistence comes from the built-in function.