BUG: apply for series of numpy arrays falis with np functions

[Aroksak]

• 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({"col": [np.array([.1, .2], dtype=float)]})
np.sin(df['col'].iloc[0])   # This works fine
df["col"].apply(np.sin)     # This breaks

Raises:

AttributeError: 'numpy.ndarray' object has no attribute 'sin'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "~/main.py", line 6, in <module>
    df["col"].apply(np.sin)
  File "~/anaconda3/envs/pandas-issue/lib/python3.7/site-packages/pandas/core/series.py", line 3840, in apply
    return f(self)
  File "~/anaconda3/envs/pandas-issue/lib/python3.7/site-packages/pandas/core/series.py", line 679, in __array_ufunc__
    result = getattr(ufunc, method)(*inputs, **kwargs)
TypeError: loop of ufunc does not support argument 0 of type numpy.ndarray which has no callable sin method

Problem description

When I have a pd.Series where each row is a numerical numpy array, I can't apply numpy functions to them, I tried np.sin, np.cos, np.exp, but I think it's a general problem. I can apply them to individual rows manually and it works fine.

Moreover, If I import tqdm with tqdm.pandas and use progress_apply instead of apply it also works just fine!

Expected Output

0    [0.09983341664682815, 0.19866933079506122]
Name: col, dtype: object

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.7.6.final.0
python-bits : 64
OS : Linux
OS-release : 5.3.0-46-generic
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 : 1.18.1
pytz : 2019.3
dateutil : 2.8.1
pip : 20.0.2
setuptools : 46.1.3.post20200325
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]

Hmm, given that this works either by wrapping the ufunc in a lambda or using map I think it likely qualifies as a bug (in the meantime you should be able to use either of these workarounds):

[ins] In [1]: ser = pd.Series([np.array([1, 2])])

[ins] In [2]: ser.map(np.sin)
Out[2]:
0    [0.8414709848078965, 0.9092974268256817]
dtype: object

[ins] In [3]: ser.apply(lambda x: np.sin(x))
Out[3]:
0    [0.8414709848078965, 0.9092974268256817]
dtype: object

[sathyz]

I wouldn't say, the failure is for all numpy functions,

>>> import pandas as pd
>>> import numpy as np
>>> df = pd.DataFrame({"col": [np.array([.1, .2], dtype=float), np.array([.3]) ]})
>>> df.col.apply(np.max)
0    0.2
1    0.3
Name: col, dtype: float64
>>>

I suppose, pandas is calling vectorised version of NumPy calls, which too raises the same exception.

>>> np.sin(df.col)
AttributeError: 'numpy.ndarray' object has no attribute 'sin'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/skmohan 1/Workspace/github.com/pandas/pandas/core/series.py", line 732, in __array_ufunc__
    result = getattr(ufunc, method)(*inputs, **kwargs)
TypeError: loop of ufunc does not support argument 0 of type numpy.ndarray which has no callable sin method
>>>