BUG: Setting the index changes dtype

[dmirecki]

• 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.

---

Note: Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

Code Sample, a copy-pastable example

import pandas as pd
import numpy as np

df = pd.DataFrame({'A': [1,2,3], 'B': [1,10,100]}).astype(np.uint16)
print(df.dtypes)
# A    uint16
# B    uint16
# dtype: object

df = df \
  .set_index('B') \
  .reset_index()
print(df.dtypes)
# B    uint64
# A    uint16
# dtype: object

Problem description

In the example above the type of column B changed from uint16 to uint64. It may cause the memory issue, so expected behavior is to not change the types implicitly.

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.6.9.final.0
python-bits : 64
OS : Linux
OS-release : 4.19.104+
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.0.3
numpy : 1.18.4
pytz : 2018.9
dateutil : 2.8.1
pip : 19.3.1
setuptools : 46.3.0
Cython : 0.29.18
pytest : 3.6.4
hypothesis : None
sphinx : 1.8.5
blosc : None
feather : 0.4.1
xlsxwriter : None
lxml.etree : 4.2.6
html5lib : 1.0.1
pymysql : None
psycopg2 : 2.7.6.1 (dt dec pq3 ext lo64)
jinja2 : 2.11.2
IPython : 5.5.0
pandas_datareader: 0.8.1
bs4 : 4.6.3
bottleneck : 1.3.2
fastparquet : None
gcsfs : None
lxml.etree : 4.2.6
matplotlib : 3.2.1
numexpr : 2.7.1
odfpy : None
openpyxl : 2.5.9
pandas_gbq : 0.11.0
pyarrow : 0.14.1
pytables : None
pytest : 3.6.4
pyxlsb : None
s3fs : 0.4.2
scipy : 1.4.1
sqlalchemy : 1.3.17
tables : 3.4.4
tabulate : 0.8.7
xarray : 0.15.1
xlrd : 1.1.0
xlwt : 1.3.0
xlsxwriter : None
numba : 0.48.0

[phofl]

I think I figured out were things go wrong here. Can someone help me out a bit to figure out, if this behavior is intentional or not? The dtype is changed at

https://github.com/pandas-dev/pandas/blob/master/pandas/core/indexes/numeric.py#L68

The function is_dtype_equal checks if the source dtype equals np.uint64 in this case in

https://github.com/pandas-dev/pandas/blob/master/pandas/core/dtypes/common.py#L608:L643

The dtype np.uint16 is cast to np.uint64 in this case, because both dtypes are not equal obviously. Is this done on purpose here, that np.uint16 and np.uint32 are cast to np.uint64 ?
If this is not done on purpose we could check if both dtypes are from the same category (np.unsignedinteger in this case).

[jreback]

see #30517 and the linked PR (closed) with 2
more issues: #27370

you are welcome to wade thru that discussion and potential resolution

closing this as duplicate