BUG: Unexpected typecast in Series.add with integer fill_value

[davidkleiven]

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

def main():
    series1 = pd.Series([1, 2, 3])
    series2 = pd.Series([4, 5])
    print(series1.dtype)
    print(series2.dtype)

    summed_series = series1.add(series2, fill_value=0)
    print(summed_series.dtype)

main()

Issue Description

When summing two series with integers the resulting series get datatype float. The result of the print statements is

int64
int64
float64

Expected Behavior

Expect that the result of summing two integer series is an integer series.

Installed Versions

INSTALLED VERSIONS

commit : 2a953cf
python : 3.11.6.final.0
python-bits : 64
OS : Linux
OS-release : 5.15.133.1-microsoft-standard-WSL2
Version : #1 SMP Thu Oct 5 21:02:42 UTC 2023
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : nb_NO.UTF-8
LANG : nb_NO.UTF-8
LOCALE : nb_NO.UTF-8

pandas : 2.1.3
numpy : 1.26.2
pytz : 2023.3.post1
dateutil : 2.8.2
setuptools : 69.0.0
pip : 23.3.1
Cython : None
pytest : 7.4.3
hypothesis : 6.90.0
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 : 1.3.7
dataframe-api-compat: None
fastparquet : None
fsspec : None
gcsfs : None
matplotlib : None
numba : None
numexpr : 2.8.7
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pyreadstat : None
pyxlsb : None
s3fs : None
scipy : 1.11.4
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
zstandard : None
tzdata : 2023.3
qtpy : None
pyqt5 : None

[paulreece]

This is because the Series are of two different lengths. When it gets to here it calls align which converts the shorter series to Float so it can be a NaN and then complete the operation between the two series.

pandas/pandas/core/series.py

Lines 5929 to 5960 in f53eb7b

	def _binop(self, other: Series, func, level=None, fill_value=None) -> Series:
	"""
	Perform generic binary operation with optional fill value.
	Parameters
	----------
	other : Series
	func : binary operator
	fill_value : float or object
	Value to substitute for NA/null values. If both Series are NA in a
	location, the result will be NA regardless of the passed fill value.
	level : int or level name, default None
	Broadcast across a level, matching Index values on the
	passed MultiIndex level.
	Returns
	-------
	Series
	"""
	this = self
	if not self.index.equals(other.index):
	this, other = self.align(other, level=level, join="outer", copy=False)
	this_vals, other_vals = ops.fill_binop(this._values, other._values, fill_value)
	with np.errstate(all="ignore"):
	result = func(this_vals, other_vals)
	name = ops.get_op_result_name(self, other)
	out = this._construct_result(result, name)
	return cast(Series, out)

ipdb> this, other = self.align(other, level=level, join="outer", copy=False)
ipdb> this
0    1
1    2
2    3
dtype: int64
ipdb> other
0    4.0
1    5.0
2    NaN
dtype: float64

A maintainer will need to chime in on whether this is actually a bug or not. Not sure how simple this would be to fix either.

In the meantime you can add astype(int) to your summing and get your desired output:

>>> summed_series = series1.add(series2, fill_value=0).astype(int)
>>> summed_series
0    5
1    7
2    3
dtype: int64

[davidkleiven]

Yes, I use astype(int). I just report it because at least when one looks at the code, I think the behavior is surprising. But there may be reasons for doing so, and if so, one could perhaps say something about it in the documentation.