loo
Efficient Leave-One-Out Cross-Validation and WAIC for Bayesian Models
Description
Efficient approximate leave-one-out cross-validation (LOO) for Bayesian models fit using Markov chain Monte Carlo, as described in Vehtari, Gelman, and Gabry (2017) <doi:10.1007/s11222-016-9696-4>. The approximation uses Pareto smoothed importance sampling (PSIS), a new procedure for regularizing importance weights. As a byproduct of the calculations, we also obtain approximate standard errors for estimated predictive errors and for the comparison of predictive errors between models. The package also provides methods for using stacking and other model weighting techniques to average Bayesian predictive distributions.
Downloads
215.7K
Last 30 days
224th
2.9M
Last 90 days
3.8M
Last year
Trend: -78.6% (30d vs prior 30d)
CRAN Check Status
Show all 14 flavors
| Flavor | Status |
|---|---|
| r-devel-linux-x86_64-debian-clang | OK |
| r-devel-linux-x86_64-debian-gcc | OK |
| r-devel-linux-x86_64-fedora-clang | OK |
| r-devel-linux-x86_64-fedora-gcc | OK |
| r-devel-macos-arm64 | OK |
| r-devel-windows-x86_64 | OK |
| r-oldrel-macos-arm64 | OK |
| r-oldrel-macos-x86_64 | OK |
| r-oldrel-windows-x86_64 | OK |
| r-patched-linux-x86_64 | OK |
| r-release-linux-x86_64 | OK |
| r-release-macos-arm64 | OK |
| r-release-macos-x86_64 | OK |
| r-release-windows-x86_64 | OK |
Check History
OK 14 OK · 0 NOTE · 0 WARNING · 0 ERROR · 0 FAILURE Mar 27, 2026
ERROR 13 OK · 0 NOTE · 0 WARNING · 1 ERROR · 0 FAILURE Mar 20, 2026
tests
Running 'testthat.R' [29s]
Running the tests in 'tests/testthat.R' failed.
Complete output:
> library(loo)
This is loo version 2.9.0
- Online documentation and vignettes at mc-stan.org/loo
- As of v2.0.0 loo defaults to 1 core but we recommen
...[truncated]...
└─base::lapply(...)
15. └─testthat (local) FUN(X[[i]], ...)
16. └─private$handle_error(msg, i)
17. └─cli::cli_abort(...)
18. └─rlang::abort(...)
Execution halted