Leave-one-out cross-validation for imugap_fit — loo.imugap_fit • imuGAP

Computes approximate leave-one-out cross-validation (LOO-CV) using Pareto smoothed importance sampling (PSIS-LOO) via the {loo} package.

Usage

# S3 method for class 'imugap_fit'
loo(x, posterior_size = NULL, ...)

Arguments

x: an imugap_fit object returned by sampling().
posterior_size: optional integer scalar; how many draws to use from the end of each chain? (default: NULL, which uses all draws).
...: additional arguments passed to loo::loo() (e.g. cores, r_eff). When r_eff is omitted and the fit contains multiple chains, relative effective sample size is automatically calculated via loo::relative_eff().

Value

for loo.imugap_fit(): an object of class loo, as returned by loo::loo().

Examples

if (FALSE) { # interactive() && requireNamespace("loo", quietly = TRUE)
# \donttest{
data("fit_sim", package = "imuGAP")
loo_res <- loo::loo(fit_sim, posterior_size = 100)
print(loo_res)
# }
}