Evaluation of posterior fit #1023
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #1023 +/- ##
==========================================
- Coverage 76.37% 76.36% -0.02%
==========================================
Files 84 84
Lines 6507 6512 +5
==========================================
+ Hits 4970 4973 +3
- Misses 1537 1539 +2
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
Ziaeemehr
changed the title
sbi/analysis/sensitivity_analysis.py
Outdated
| @@ -501,3 +502,53 @@ def project(self, theta: Tensor, num_dimensions: int) -> Tensor: | |||
| projected_theta = torch.mm(theta, projection_mat) | |||
|
|
|||
| return projected_theta | |||
|
|
|||
|
|
|||
| def posterior_shrinkage(prior_std, post_std): | |||
There was a problem hiding this comment.
Currently, the function does not do much work.
Maybe change inputs to prior and posterior samples and do the estimation of the standard deviation within?
There was a problem hiding this comment.
Type hints should also be here and not in the docstring.
sbi/analysis/sensitivity_analysis.py
Outdated
|
|
||
| Parameters | ||
| ---------- | ||
| prior_std : float, array-like |
There was a problem hiding this comment.
Typing should also be valid for tensors (i.e. vectors)
I guess one can define this metric also in a multivariate setting, just with the "total stddev" i.e. the sum.
sbi/analysis/sensitivity_analysis.py
Outdated
| ---------- | ||
| prior_std : float, array-like | ||
| The standard deviation of the prior distribution. | ||
| post_std : float, array-like |
sbi/analysis/sensitivity_analysis.py
Outdated
| return 1 - (post_std / prior_std) ** 2 | ||
|
|
||
|
|
||
| def posterior_zscore(true_mean, post_mean, post_std): |
There was a problem hiding this comment.
Same as above i.e. type hints here.
And the metric should be suitable for vector input.
|
@manuelgloeckler Thanks for the comments. |
Add to function for evaluation of posterior fit.
Fixes issue #1017
Checklist
Put an
xin the boxes that apply. You can also fill these out after creatingthe PR. If you're unsure about any of them, don't hesitate to ask. We're here to
help! This is simply a reminder of what we are going to look for before merging
your code.
guidelines
with
pytest.mark.slow.guidelines
main(or there are no conflicts withmain)