-
-
Notifications
You must be signed in to change notification settings - Fork 34
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
improved E_loo Pareto-k diagnostics #247
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #247 +/- ##
==========================================
+ Coverage 92.44% 92.78% +0.34%
==========================================
Files 31 31
Lines 2831 2829 -2
==========================================
+ Hits 2617 2625 +8
+ Misses 214 204 -10 ☔ View full report in Codecov by Sentry. |
This looks good, but I'm still not sure what to do about the posterior dependency. If we keep it like this then we should move it to Imports from Suggests, but I think my slight preference would be to copy anything needed over to loo, but just for right now. Then eventually we could do a bigger update where we convert to using posterior everywhere in the loo package where it makes sense (and then delete the copied code and add a strong dependency on posterior). For example, I think we can also use |
Option 1. is easy as I already tested it (by keeping copying missing functions until it did work), but is 400 lines too much? |
I didn’t realize it would be that much code. Thats a good reason for just adding the dependency on posterior since we probably don’t want to maintain a lot of duplicate code in different packages. But I’ll think about it a little more today before we make a final decision. |
It's about 200 lines for Pareto diagnostic and smoothing which supports both non-logged and logged inputs and both tails, and about 200 lines for ess_tail and helper functions. I guess a simple version which would use only functions already in |
Ok I think let's just go ahead and put posterior in Imports. Since many of our other packages already import (or will import) posterior I guess it's not such a big deal for loo to import it too. Then we can also remove some code in |
Ok everything is passing so I will merge this and then start running reverse dependency checks again |
Reverse dependency checks passed except for the one package we already knew about. I submitted a PR to fix that package almost a month ago (and it was merged already), so I will go ahead and submit loo to CRAN. |
On CRAN now |
Fixes #236
posterior::pareto_khat
, but this can be copied toloo
if we want to avoid the dependency