Deprecate run_ica()

[hoechenberger]

What does this implement/fix?

Removes the mne.preprocessing.ica.run_ica() function. All it does it instantiate an ICA object and then run an artifact detection. However, the same can be achieved by using ICA.detect_artifacts().
Deprecates mne.preprocessing.ica.run_ica() in favor of ICA.detect_artifacts()

[codecov]

Codecov Report

Merging #7342 into master will decrease coverage by 0.05%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #7342      +/-   ##
==========================================
- Coverage   89.99%   89.94%   -0.06%     
==========================================
  Files         450      450              
  Lines       80659    81587     +928     
  Branches    12943    13003      +60     
==========================================
+ Hits        72589    73382     +793     
- Misses       5244     5385     +141     
+ Partials     2826     2820       -6     

[larsoner]

git grep run_ica tells me that examples/preprocessing/plot_ica_comparison.py should be updated, too

[hoechenberger]

git grep run_ica tells me that examples/preprocessing/plot_ica_comparison.py should be updated, too

No, it defines its own run_ica function which it then calls with different parameter values to create different plots

[larsoner]

Hah! fair enough

[hoechenberger]

@larsoner @drammock I believe I've addressed all of your comments.

[larsoner]

LGTM +1 for merge once CIs come back happy

[dengemann]

Good call! run_ica has been some historical leftover for years ...

…

On 19 Feb 2020, at 21:23, Eric Larson ***@***.***> wrote:  @larsoner approved this pull request. LGTM +1 for merge once CIs come back happy — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[drammock]

The Travis failure is a general timeout error at the CodeCov step, seems unrelated to this PR (@larsoner have you seen that before? Seems odd). Azure is "Windows fatal exception: caccess violation" which is also unrelated (though is happening a lot lately, should be investigated).

[larsoner]

@drammock the Travis timeouts at various stages are just numpy/numpy#15580

The Azure failures I was hoping to fix with #7295 but hasn't quite done the trick