BUG: comparison warning with NaN when DataFrame/Series broadcasting

[Zhang18]

Code Sample

>>> df = pd.DataFrame({'A' : [np.nan, 2.0, np.nan]})
>>> df
     A
0  NaN
1  2.0
2  NaN
>>> x = pd.Series([1, 1, 1])
>>> x
0    1
1    1
2    1
dtype: int64
>>> df.clip_lower(x, axis=0)
/usr/local/lib/python3.5/dist-packages/pandas/core/ops.py:1253: RuntimeWarning: invalid value encountered in greater_equal
  result = op(x, y)
     A
0  NaN
1  2.0
2  NaN

>>> df.clip_lower(x, axis=0)
     A
0  NaN
1  2.0
2  NaN

Problem description

So the RuntimeWarning is only thrown the first time the clip is called. But not on subsequent times. This is inconsistent.

Expected Output

I propose to suppress this warning altogether as the triggering use cases seem to be completely common and fully legit.

Output of pd.show_versions()

>>> pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.5.2.final.0
python-bits: 64
OS: Linux
OS-release: 4.4.0-75-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8

pandas: 0.20.1
pytest: None
pip: 9.0.1
setuptools: 20.7.0
Cython: None
numpy: 1.12.1
scipy: 0.19.0
xarray: None
IPython: 6.0.0
sphinx: None
patsy: 0.4.1
dateutil: 2.6.0
pytz: 2017.2
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: 2.0.1
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: 0.7.3
lxml: None
bs4: None
html5lib: 0.999999999
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.8
s3fs: None
pandas_gbq: None
pandas_datareader: None

[TomAugspurger]

This repeatedly shows warnings for me. Do you perhaps set a warningfilter somewhere like

import warnings
warnings.simplefilter('once', RuntimeWarning)

an easy way to test is to call

In [19]: def f():
    ...:     raise RuntimeWarning("test")
    ...:     return 2

a couple times.

[Zhang18]

Here is a continuous screen output per your suggestion. Apparently f() keeps raise warnings whereas clip doesn't.

Python 3.5.2 (default, Nov 17 2016, 17:05:23)
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> def f():
...     raise RuntimeWarning("test")
...     return 1
...
>>> f()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 2, in f
RuntimeWarning: test
>>> f()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 2, in f
RuntimeWarning: test
>>> f()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 2, in f
RuntimeWarning: test
>>> df = pd.DataFrame({'A' : [np.nan, 2.0, np.nan]})
>>> x = pd.Series([1, 1, 1])
>>> df.clip_lower(x, axis=0)
/usr/local/lib/python3.5/dist-packages/pandas/core/ops.py:1253: RuntimeWarning: invalid value encountered in greater_equal
  result = op(x, y)
     A
0  NaN
1  2.0
2  NaN
>>> df.clip_lower(x, axis=0)
     A
0  NaN
1  2.0
2  NaN

[TomAugspurger]

Can you post the output of warnings.filters? It seems unlikely, but it's possible to filter warnings form specific libraries.

[Zhang18]

Here you go:

>>> pprint.pprint(warnings.filters)
[('ignore',
  re.compile('numpy.ndarray size changed', re.IGNORECASE),
  <class 'Warning'>,
  re.compile(''),
  0),
 ('ignore',
  re.compile('numpy.ufunc size changed', re.IGNORECASE),
  <class 'Warning'>,
  re.compile(''),
  0),
 ('ignore',
  re.compile('numpy.dtype size changed', re.IGNORECASE),
  <class 'Warning'>,
  re.compile(''),
  0),
 ('always', None, <class 'numpy.lib.polynomial.RankWarning'>, None, 0),
 ('default', None, <class 'urllib3.exceptions.SNIMissingWarning'>, None, 0),
 ('default', None, <class 'urllib3.exceptions.SubjectAltNameWarning'>, None, 0),
 ('ignore', None, <class 'DeprecationWarning'>, None, 0),
 ('ignore', None, <class 'PendingDeprecationWarning'>, None, 0),
 ('ignore', None, <class 'ImportWarning'>, None, 0),
 ('ignore', None, <class 'BytesWarning'>, None, 0),
 ('ignore', None, <class 'ResourceWarning'>, None, 0),
 ('always', None, <class 'urllib3.exceptions.SecurityWarning'>, None, 0),
 ('default',
  None,
  <class 'urllib3.exceptions.InsecurePlatformWarning'>,
  None,
  0),
 ('default', None, <class 'requests.exceptions.FileModeWarning'>, None, 0)]

[Zhang18]

But inconsistency aside, do you agree this warning is unnecessary and should be removed?

[TomAugspurger]

Sorry, missed that section of your original post. I think that was fixed by #16373 Can you try on master?

[jreback]

This is a different issue.
This has to do with broadcasting of a Series and a DataFrame.

In [6]: df = pd.DataFrame({'A' : [np.nan, 2.0, np.nan]})

In [7]: x = pd.Series([1, 1, 1])

In [8]: df.ge(x, axis=0)
/Users/jreback/pandas/pandas/core/ops.py:1253: RuntimeWarning: invalid value encountered in greater_equal
  result = op(x, y)
Out[8]: 
       A
0  False
1   True
2  False

[jreback]

I recall the same issue, but can't find it w/o a quick search. @TomAugspurger any idea?

[jreback]

note that this fix is the same as in #16373, e.g. wrapping with np.errstate

[andrewarcher]

I'm going to start working on this.