BUG: pd.DataFrame is created inconsistently from numpy.ndarray for non-native pandas dtypes such as: datetime64[ms], etc..

[mheydel]

• 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

a = np.array(
    [
        ["2015-01-01", "2015-01-02", "2015-01-03"],
        ["2017-01-01", "2017-01-02", "2017-02-03"],
    ],
    dtype="datetime64[ms]",
)

print(pd.DataFrame(a))
           0          1          2
0 2015-01-01 2015-01-03 2017-01-02
1 2015-01-02 2017-01-01 2017-02-03


b = np.array(
    [
        ["2015-01-01", "2015-01-02", "2015-01-03"],
        ["2017-01-01", "2017-01-02", "2017-02-03"],
    ],
    dtype="datetime64[M]",
)

print(pd.DataFrame(b))
           0          1          2
0 2015-01-01 2015-01-03 2017-01-02
1 2015-01-02 2017-01-01 2017-02-03

c = np.array(
    [
        ["2015-01-01", "2015-01-02", "2015-01-03"],
        ["2017-01-01", "2017-01-02", "2017-02-03"],
    ],
    dtype="datetime64[ns]",
)

print(pd.DataFrame(c))
           0          1          2
0 2015-01-01 2015-01-02 2015-01-03
1 2017-01-01 2017-01-02 2017-02-03

Problem description

The exists an inconsistency how pd.DataFrame() is created from 2D numpy.ndarray for not pandas native type. The dataframe should be transposed for datetime64[M], datetime64[ms] etc.

Expected Output

print(pd.DataFrame(a))
           0          1          2
0 2015-01-01 2015-01-02 2015-01-03
1 2017-01-01 2017-01-02 2017-02-03

print(pd.DataFrame(b))
0 2015-01-01 2015-01-02 2015-01-03
1 2017-01-01 2017-01-02 2017-02-03

Possible Hint

d = np.array(
    [
        ["2015-01-01", "2015-01-02", "2015-01-03"],
        ["2017-01-01", "2017-01-02", "2017-02-03"],
    ],
    dtype="datetime64[ms]",
    order="f",
)

print(pd.DataFrame(d))
0 2015-01-01 2015-01-02 2015-01-03
1 2017-01-01 2017-01-02 2017-02-03

Output of pd.show_versions()

INSTALLED VERSIONS

commit : db08276
python : 3.8.5.final.0
python-bits : 64
OS : Windows
OS-release : 10
Version : 10.0.19041
machine : AMD64
processor : Intel64 Family 6 Model 158 Stepping 9, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : English_United Kingdom.1252

pandas : 1.1.3
numpy : 1.19.2
pytz : 2020.1
dateutil : 2.8.1
pip : 21.0
setuptools : 49.6.0.post20210108
Cython : None
pytest : 6.2.1
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.11.2
IPython : 7.18.1
pandas_datareader: None
bs4 : 4.9.3
bottleneck : None
fsspec : None
fastparquet : None
gcsfs : None
matplotlib : 3.3.3
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pytables : None
pyxlsb : None
s3fs : None
scipy : 1.6.0
sqlalchemy : 1.3.22
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
numba : None

[jorisvandenbossche]

This was actually a regression (compared to 1.0.5, but still worth including in the bug fix release IMO since it's silently giving garbled results)

[jorisvandenbossche]

Backported in #39475