Parameter ensemble plot #1574
Parameter ensemble plot #1574
Conversation
plakrisenko
requested review from
Doresic,
PaulJonasJost,
stephanmg and
vwiela
as code owners
pypesto/visualize/ensemble.py
Outdated
| if ax is None: | ||
| fig, ax = plt.subplots(figsize=size) | ||
|
|
||
| x = -0.4 |
There was a problem hiding this comment.
the x coordinate for the rectangle vertex, changed now
|
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## develop #1574 +/- ##
========================================
Coverage 83.65% 83.66%
========================================
Files 163 163
Lines 14071 14101 +30
========================================
+ Hits 11771 11797 +26
- Misses 2300 2304 +4 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
stephanmg
left a comment
stephanmg
left a comment
There was a problem hiding this comment.
See single line comments.
PaulJonasJost
left a comment
PaulJonasJost
left a comment
There was a problem hiding this comment.
Thanks for this :) looks good to me
| @close_fig | ||
| def test_ensemble_parameters_plot(): | ||
| # creates a test problem | ||
| problem = create_problem(n_parameters=100) |
There was a problem hiding this comment.
such a big one necessary for the test? 🙈
Doresic
left a comment
Doresic
left a comment
There was a problem hiding this comment.
Looks really nice! Will use myself :) I have a couple of comments
- can we have a general legend on all the elements of the figure (the rectangles, red point, dashed line)
- can we have a bar on the side or at the top of the plot that shows the grey and mentions what the grey gradient actually represents? Btw, what does it actually represent? The obj. func. value of the point?
| ensemble: | ||
| ensemble of parameter vectors (from pypesto.ensemble). |
There was a problem hiding this comment.
So this is a method to generally just visualize the ensemble, so we cannot really say that the rectangles represent, for example, 95% confidence intervals. But one would need to create an ensemble in some way, and then subset it using a xi^2 threshold to get confidence intervals, right?
Would it be good to have this showcased somewhere, like in a notebook that mentiones ensembles or identifiability? How to use this visualization correctly? Or is it ok to just expand the description of the visualization in the docstring to describe that it will just visualize all members of the ensemble and that these do not represent CIs.
Visualization of parameter values included in the ensemble.