BUG: pd.Series.replace(to_replace=...) does not preserve the original dtype of the series

[chrispe]

• 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.

---

Code Sample, a copy-pastable example

>>> s = pd.Series(["one", "two"], dtype="string")
>>> expected = pd.Series(["1", "2"], dtype="string")
>>> result = s.replace(to_replace={"one": "1", "two": "2"})
>>> expected
0    1
1    2
dtype: string
>>> result
0    1
1    2
dtype: object

Problem description

The pandas.Series.replace does not return a series with the same dtype as the the original series. This behaviour is produced only when the replace function is called using the argument to_replace.

Notice: Similar behaviour is produced with other dtypes as well, this issue is not limited to string dtype.

The investigation should probably start here: /pandas/core/generic.py

Expected Output

We would expect the pandas.Series.replace(to_replace=...) to preserve the dtype of the series.
In the above code sample, both of the series should have a dtype of string.

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None python : 3.8.2.final.0 python-bits : 64 OS : Darwin OS-release : 19.4.0 machine : x86_64 processor : i386 byteorder : little LC_ALL : None LANG : None LOCALE : None.UTF-8

pandas : 1.0.3
numpy : 1.18.1
pytz : 2019.3
dateutil : 2.8.1
pip : 20.0.2
setuptools : 46.1.3.post20200330
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
fastparquet : None
gcsfs : None
lxml.etree : None
matplotlib : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pytables : None
pytest : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
xlsxwriter : None
numba : None

[dsaxton]

I think that this is not only limited to string dtype:

[ins] In [7]: s                                                                                                                                                                                              
Out[7]: 
0    1
1    2
dtype: Int64

[ins] In [8]: s.replace({1: 3})                                                                                                                                                                              
Out[8]: 
0    3
1    2
dtype: int64

[ins] In [12]: s                                                                                                                                                                                             
Out[12]: 
0     True
1    False
dtype: boolean

[ins] In [13]: s.replace({True: False})                                                                                                                                                                      
Out[13]: 
0    False
1    False
dtype: object

Also related: #32988

[chrispe]

I think that this is not only limited to string dtype:

[ins] In [7]: s                                                                                                                                                                                              
Out[7]: 
0    1
1    2
dtype: Int64

[ins] In [8]: s.replace({1: 3})                                                                                                                                                                              
Out[8]: 
0    3
1    2
dtype: int64

[ins] In [12]: s                                                                                                                                                                                             
Out[12]: 
0     True
1    False
dtype: boolean

[ins] In [13]: s.replace({True: False})                                                                                                                                                                      
Out[13]: 
0    False
1    False
dtype: object

Also related: #32988

Added this to the description, so it should be more clear now. Thanks for the feedback @dsaxton.

[simonjayhawkins]

Also related: #32988

#32988 is a regression from 0.25.3 whereas this issue was already present in 0.25.3 for Int64 type

>>> import pandas as pd
>>>
>>> pd.__version__
'0.25.3'
>>>
>>> s = pd.Series([1, 2], dtype="Int64")
>>> s
0    1
1    2
dtype: Int64
>>>
>>> s.replace(to_replace={1: 3})
0    3
1    2
dtype: int64
>>>

[MarcusJellinghaus]

I think I have the same issue:

import pandas as pd
import numpy as np
pd.__version__  #1.1.2

s = pd.Series(['abc', None, pd.NA, 'def'], dtype="string")
print(s.dtype)  # string
t = pd.Series([None], dtype="string")
print(t.dtype)  # string
u = pd.Series([pd.NA], dtype="string")
print(u.dtype)  # string

v = pd.Series(['abc', None, pd.NA], dtype="string")
print(v.dtype)  # string
w = v.replace('abc', pd.NA)
print(w.dtype) # Now of type object, why not string?
               # In Python 1.0.3, it was string