忽略来自Python模块（seaborn，sklearn）的警告

Question

There are many questions related to the question title above and all basically tell you to do:

import warnings
warnings.filterwarnings('ignore')

and to make sure this is placed before the first import.

However, even after doing this I get many warnings from seaborn and sklearn . I get UserWarning , DataConversionWarning and RuntimeWarning which, according to documentation, all inherit from Warning and should be covered by the above code.

Is there another way to hide those warnings? (I cannot really solve most of them anyway)

EDIT

Example 1:

C:\Anaconda3\lib\site-packages\sklearn\preprocessing\data.py:645: DataConversionWarning: Data with input dtype int32, int64 were all converted to float64 by StandardScaler.
  return self.partial_fit(X, y)

Example 2

C:\Anaconda3\lib\site-packages\seaborn\distributions.py:340: UserWarning: Attempted to set non-positive bottom ylim on a log-scaled axis.
Invalid limit will be ignored.
  ax.set_ylim(0, auto=None)

Answer 1

Example2

It's a bit hard to track down; seaborn imports statsmodels. And in statsmodels/tools/sm_exceptions.py you find this line

warnings.simplefilter('always', category=UserWarning)

in which reverses any previous setting for user warnings.

A solution for now would be to remove that line or to set the warning state after the import of seaborn (and hence statsmodels). In a future version of statsmodels this will be fixed by PR 4712 , so using the development version of statsmodels would also be an option.

Example1

I did not find a way to reproduce the first example from sklearn ; so that may or may not have a different reason.