BUG: Pandas changes dtypes of columns when no float (or other) assignments are done to this column

[volsend]

df = pd.DataFrame()
df["a"] = int(0)

for col_name in ["b", "c"]:
    df[col_name] = 0.

print(df.dtypes)
    
for idx, b in enumerate([1,2,3]):
    df.loc[df.shape[0]] = {
        'a' : int(idx),
        'b' : float(b),
        'c' : float(b),
    }

print(df.dtypes)

Problem description

In the above example, pandas changes the dtype of the column a of dataframe df from int64 to float64, while there is no operation, that assigns non-integer value to this column.

Expected Output

Currently, the output of this code is:

a      int64
b    float64
c    float64
dtype: object
a    float64
b    float64
c    float64
dtype: object

while the expected one is:

a      int64
b    float64
c    float64
dtype: object
a      int64
b    float64
c    float64
dtype: object

pd.show_versions() output

INSTALLED VERSIONS
------------------
commit           : None
python           : 3.6.9.final.0
python-bits      : 64
OS               : Linux
OS-release       : 4.15.0-101-generic
machine          : x86_64
processor        : x86_64
byteorder        : little
LC_ALL           : None
LANG             : C.UTF-8
LOCALE           : en_US.UTF-8

pandas           : 1.0.4
numpy            : 1.18.1
pytz             : 2019.3
dateutil         : 2.8.1
pip              : 20.1.1
setuptools       : 44.0.0
Cython           : None
pytest           : None
hypothesis       : None
sphinx           : None
blosc            : None
feather          : None
xlsxwriter       : None
lxml.etree       : 4.5.0
html5lib         : None
pymysql          : None
psycopg2         : None
jinja2           : 2.10.3
IPython          : 7.11.1
pandas_datareader: None
bs4              : None
bottleneck       : None
fastparquet      : None
gcsfs            : None
lxml.etree       : 4.5.0
matplotlib       : 3.1.2
numexpr          : None
odfpy            : None
openpyxl         : None
pandas_gbq       : None
pyarrow          : None
pytables         : None
pytest           : None
pyxlsb           : None
s3fs             : None
scipy            : 1.4.1
sqlalchemy       : None
tables           : None
tabulate         : None
xarray           : None
xlrd             : 1.2.0
xlwt             : None
xlsxwriter       : None
numba            : None

[jorisvandenbossche]

My guess is that we might convert the dictionary into a Series first (which would turn it into a float series). Just a guess though, didn't actually check.