Correlation plot with nans

[PaulJonasJost]

Added a test. There was previously an error, when "all" start indices were selected, where no general checks were run.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 77.77778% with 2 lines in your changes are missing coverage. Please review.

Project coverage is 84.51%. Comparing base (3edf785) to head (dcb1a06).

Files	Patch %	Lines
pypesto/visualize/misc.py	77.77%	2 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1365      +/-   ##
===========================================
+ Coverage    84.42%   84.51%   +0.08%     
===========================================
  Files          157      157              
  Lines        12928    12931       +3     
===========================================
+ Hits         10915    10928      +13     
+ Misses        2013     2003      -10     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[stephanmg]

LGTM.

[Doresic]

The process_start_indices function is used by multiple visualization routines. And some of them also have specific ways of dealing with infinite values. I think those should then be updated as well accordingly if process_start_indices doesn't return start indices with infinite values anymore. At least removed, so that the start indices are not checked for infinite values.

We should also think about adding a warning or logging entry that some start indices were removed from plotting due to infinite values (if that's the case). That's, for instance, done in the parameters.py plotting routine.

[Doresic]

Should we check for np.isnan here as well?

[stephanmg]

np.isfinite tests for Not a Number already, in case of NaN returns False.

[stephanmg]

We should also think about adding a warning or logging entry that some start indices were removed from plotting due to infinite values (if that's the case). That's, for instance, done in the parameters.py plotting routine.

Could that be as simple as a string message / status with: "Removed XY entries due to infinite or NaN value"?

[m-philipps]

Looks good to me, I would update the process_start_indices description, e.g.:

Create an array of indices if a number was provided, checks that the
indices do not exceed the max_index and removes starts with non-finite fval.