Refactor sample_posterior_predictive_w for V4
#5800
Conversation
|
fast posterior_predictive_sampling no longer exists (everything is fast now). Also some tweaks will be needed following #5787 which should be merged soon! |
Codecov Report
@@ Coverage Diff @@
## main #5800 +/- ##
==========================================
+ Coverage 88.99% 89.43% +0.43%
==========================================
Files 74 74
Lines 13809 13807 -2
==========================================
+ Hits 12290 12348 +58
+ Misses 1519 1459 -60
|
|
Let me make those tweaks and push tonight. |
| @@ -1958,13 +1958,14 @@ def sample_posterior_predictive_w( | |||
| # TODO sample_posterior_predictive_w is currently only work for model with | |||
| # one observed. | |||
| # XXX: This needs to be refactored | |||
| # ppc[var.name].append(draw_values([var], point=param, size=size[idx])[0]) | |||
| raise NotImplementedError() | |||
| ppcl[var.name].append(draw([var])[0]) | |||
There was a problem hiding this comment.
This won't be conditioned on the posterior samples.
You will have to create a custom function similarly to what sample_posterior_predictive does. See recent changes in #5759
There was a problem hiding this comment.
It might also be useful to compare what changed between V3 and V4 in sample_posterior_predictive because some of the things in sample_posterior_predictive_w might no longer be needed / make sense
zaxtax
force-pushed
the
refactor-sample_posterior_predictive_w
branch
from
5bb0f63 to
f1d0f91
Compare
zaxtax
force-pushed
the
refactor-sample_posterior_predictive_w
branch
from
f1d0f91 to
6683916
Compare
ricardoV94
changed the title
sample_posterior_predictive_w for V4sample_posterior_predictive_w for V4
|
Closing in favour of #6254 |
This tries to address #4807
Right now, it passes all unit tests, but doesn't use the fast sampling procedure of
sample_posterior_predictiveThe code is also not feature-complete.