add animations to plot_ppc #546
Conversation
1d36aed to
47a2d9f
Compare
|
Is there a way to increase number of samples from 1 to N? e.g. 3 |
arviz/plots/ppcplot.py
Outdated
| animate, | ||
| np.arange(0, num_pp_samples), | ||
| init_func=init, | ||
| blit=True, |
There was a problem hiding this comment.
This will break stuff (at least it did on osx) and is probably not needed if animation is saved to a file.
There was a problem hiding this comment.
Not clear to me what will break stuff. The idea is to be able to work with animations inside a Jupyter notebook, not just save them to a file.
There was a problem hiding this comment.
There was a problem hiding this comment.
So, what we can do here, add a keyword so users can set the value of blit?, or we detect the OS and if it is osx we use TkAgg? (BTW I don't have access to a machine with osx)
There was a problem hiding this comment.
I think let users decide. Maybe throw warning on osx?
There was a problem hiding this comment.
Do you like something like "If you experience problems rendering the animation try turning blit to False or setting the plotting backend to TkAgg".
There was a problem hiding this comment.
or better "If you experience problems rendering the animation try setting `animation_kwargs({'blit':False}) or setting the plotting backend to TkAgg."
|
Do you mean to render N samples per frame? |
|
Yes. Like having 3 to 10 samples per frame. |
|
Not with the current implementation but it can be added. A friend of mine suggested to have an option to pile-up samples, probably with a lower value of alpha. My concern with these options is if they goes against the purpose of the animation, that is to minimize "deterministic construal errors", but honestly I have to learn more about this and about Hypothetical Outcome Plots. |
|
@ahartikainen If you don't mind I prefer to left the N-samples per frame for another PR, after I have the time to do some more reading. I could open an issue. |
|
@ahartikainen do you think this is ready to merge? (and further improve in the future) |
|
Yes, LGTM, sorry I have missed your question before. |
Some examples
kind=densitykind=scatterkind=cumulativekind=densitydiscrete variables