loo
loo copied to clipboard
stan-dev
loo R package for approximate leave-one-out cross-validation (LOO-CV) and Pareto smoothed importance sampling (PSIS)
- Build site in master to `.\docs` - GitHub Action builds site on releases (double check to ensure Action doesn't wipe root and replace with static files) - Updated favicons...
Use Posit's new opinionated formatter, [Air](https://posit-dev.github.io/air/formatter.html) to [review pull requests](https://posit-dev.github.io/air/integration-github-actions.html#format-with-github-suggestions) to ensure they conform to the style as well as format the entire project once. What do you think? @jgabry...
Rely on `posterior` for pareto smoothening tails. Some tests should again fail due to minor numerical differences. Need to bump posterior dependency on next release.
This is related to issues #223, #220, #213, #201, #135, #106 We have discussed rewriting CRPS and SCRPS functions to use better computation (probability weighted moment form). As new functions...
Bumps [codecov/codecov-action](https://github.com/codecov/codecov-action) from 4 to 5. Release notes Sourced from codecov/codecov-action's releases. v5.0.0 v5 Release v5 of the Codecov GitHub Action will use the Codecov Wrapper to encapsulate the CLI....
![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot] avatar"){kind=avatar}
The corresponding issue #299 This PR is not ready for merge, but needs some thought
Look into where we use parallelization and see which instances rely on forking for major performance benefits.
Since we do use `posterior` package elsewhere, replace gpdfit() and qgpd() with posterior::gpdfit and posterior::qgeneralized_pareto() This will make maintenance easier. For example, there is a PR for posterior::gpdfit() to make...
since gpdfit is not exported from posterior, copy the latest version from posterior
Paper [Uncertainty in Bayesian Leave-One-Out Cross-Validation Based Model Comparison](https://arxiv.org/abs/2008.10296) describes when `elpd_diff` and `se_diff` based normal approximation can be expected to be well calibrated. The paper includes case studies with...