Efficient Leave-One-Out Cross-Validation and WAIC for Bayesian Models
Thin a draws object
Compute weighted expectations
Extract importance sampling weights
The number of posterior draws in a draws object.
Pareto smoothed importance sampling (PSIS) using approximate posterior...
Model comparison (deprecated, old version)
Continuously ranked probability score
Compute a point estimate from a draws object
Generic (expected) log-predictive density
Objects to use in examples and tests
Extract pointwise log-likelihood from a Stan model
Find the model names associated with "loo" objects
Estimate parameters of the Generalized Pareto distribution
Efficient approximate leave-one-out cross-validation (LOO) using subsa...
A parent class for different importance sampling methods.
Generic function for K-fold cross-validation for developers
Helper functions for K-fold cross-validation
Efficient approximate leave-one-out cross-validation (LOO) for posteri...
Model comparison
Model averaging/weighting via stacking or pseudo-BMA weighting
Split moment matching for efficient approximate leave-one-out cross-va...
Moment matching for efficient approximate leave-one-out cross-validati...
Estimate leave-one-out predictive performance..
Convenience function for computing relative efficiencies
Datasets for loo examples and vignettes
LOO package glossary
Efficient LOO-CV and WAIC for Bayesian models
Efficient approximate leave-one-out cross-validation (LOO)
Named lists
The number of observations in a psis_loo_ss object.
Get observation indices used in subsampling
Extractor methods
Standard importance sampling (SIS)
Truncated importance sampling (TIS)
Update psis_loo_ss objects
Widely applicable information criterion (WAIC)
Parallel psis list computations
Diagnostics for Pareto smoothed importance sampling (PSIS)
Convenience function for extracting pointwise estimates
Print dimensions of log-likelihood or log-weights matrix
Print methods
Diagnostics for Laplace and ADVI approximations and Laplace-loo and AD...
Pareto smoothed importance sampling (PSIS)
Pareto smoothed importance sampling (deprecated, old version)
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.
Useful links