Description
-
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 master branch of pandas.
Reproducible Example
import pandas as pd
df = pd.DataFrame({'idx': [0, 1, 0, 1],
'col': list('aabb'),
'ints': [10, 20, 30, 40],
'ints2': [11, 21, 31, 41],
'floats': [0.1, 0.2, 0.3, 0.4],
'strs': list('ABCD'),
'dts': pd.date_range('2021-01-01 10:00:00', '2021-01-04 10:00:00').to_series()})
df.pivot(index='idx', columns='col', values=['ints', 'floats'])
# ints floats
# col a b a b
# idx
# 0 10.0 30.0 0.1 0.3
# 1 20.0 40.0 0.2 0.4
df.pivot(index='idx', columns='col', values=['ints', 'ints2']).dtypes # -> all int
# col
# ints a int64
# b int64
# ints2 a int64
# b int64
# dtype: object
df.pivot(index='idx', columns='col', values=['ints', 'floats']).dtypes # -> all float
# col
# ints a float64
# b float64
# floats a float64
# b float64
# dtype: object
df.pivot(index='idx', columns='col', values=['ints', 'strs']).dtypes # -> all object
# col
# ints a object
# b object
# strs a object
# b object
# dtype: object
df.pivot(index='idx', columns='col', values=['ints', 'dts']).dtypes # -> object and datetime64
# col
# ints a object
# b object
# dts a datetime64[ns]
# b datetime64[ns]
# dtype: objectIssue Description
When calling DataFrame.pivot with a list of column names as the values argument, numeric columns are cast to a common ancestor datatype of the selected columns. So depending on the dtypes of the columns selected in the values argument:
- If only int columns are selected, the dtype is maintained
- If at least a float column is selected, all int columns are cast to float.
- If at least an object or datetime column is selected, all numeric columns (int and floats) are cast to object
- datetime columns remain as datetime column even if some object column is selected
- Object columns remain as object dtype
Expected Behavior
As far as I understand, the dtype of pivoted columns could be maintained. If there are missing values on the pivot table, int columns might be casted to floats to allow NaNs, but otherwise no dtype transformations should occur due to pivot.
Perhaps, the behaviour is expected as if one were to do df.stack().unstack(). In that case, perhaps it could be included in the notes section of the documentation.
Installed Versions
INSTALLED VERSIONS
commit : c7f7443
python : 3.9.2.final.0
python-bits : 64
OS : Darwin
OS-release : 20.6.0
Version : Darwin Kernel Version 20.6.0: Wed Jun 23 00:26:31 PDT 2021; root:xnu-7195.141.2~5/RELEASE_X86_64
machine : x86_64
processor : i386
byteorder : little
LC_ALL : None
LANG : None
LOCALE : None.UTF-8
pandas : 1.3.1
numpy : 1.21.1
pytz : 2021.1
dateutil : 2.8.2
pip : 21.2.3
setuptools : 57.4.0
Cython : None
pytest : 6.2.4
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : 2.9.1 (dt dec pq3 ext lo64)
jinja2 : 3.0.1
IPython : 7.26.0
pandas_datareader: None
bs4 : None
bottleneck : None
fsspec : None
fastparquet : None
gcsfs : None
matplotlib : 3.4.2
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pyxlsb : None
s3fs : None
scipy : 1.7.1
sqlalchemy : None
tables : None
tabulate : 0.8.9
xarray : None
xlrd : None
xlwt : None
numba : None