Approximate leave-one-out cross-validation for AME / LAME fits

S3 method for loo that uses the per-iteration pointwise log-likelihood stored on the fit object (fit$log_lik) when the model was fit with save_log_lik = TRUE. Returns the standard loo object with Pareto-k diagnostics.

Usage

# S3 method for class 'ame'
loo(x, ...)

# S3 method for class 'lame'
loo(x, ...)

# S3 method for class 'ame_als'
loo(x, ...)

Arguments

x

A fitted ame or lame object that has $log_lik.

...

Additional arguments forwarded to loo::loo.matrix (e.g. cores, r_eff).

Value

A loo object.

Details

What log_lik measures. For family in {normal, binary, cbin, poisson, ordinal} the stored pointwise log-likelihood is the exact family-specific Y density on the response scale, so elpd_loo is directly comparable to a loo() output from Stan / brms fit to the same family. For the rank likelihood frn the exact marginal needs GHK Monte Carlo (Halton sequence); on the longitudinal lame() path you can opt in with log_lik_method = "observed_ghk", on the cross-sectional ame() path the fallback is the augmented-Z normal approximation (with a one-time warning). Inspect fit$log_lik_method on any fit to see which branch was used.

Chunked log-lik portability. When fit with save_log_lik = "chunked", the on-disk chunk files default to tempdir(), which is cleared at the end of the R session. If you intend to saveRDS() the fit and reload it in a fresh session, supply an explicit persistent log_lik_path (e.g. "./loglik_chunks") so the chunks survive the round trip.

Examples

# \donttest{
data(YX_nrm)
fit <- ame(YX_nrm$Y, Xdyad = YX_nrm$X, R = 0,
           nscan = 60, burn = 15, odens = 5,
           save_log_lik = TRUE, verbose = FALSE)
if (requireNamespace("loo", quietly = TRUE)) {
  loo_res <- loo::loo(fit)
  print(loo_res)
}
#> Warning: Not enough tail samples to fit the generalized Pareto distribution in some or all columns of matrix of log importance ratios. Skipping the following columns: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, ... [9890 more not printed].
#> Warning: Some Pareto k diagnostic values are too high. See help('pareto-k-diagnostic') for details.
#> 
#> Computed from 12 by 9900 log-likelihood matrix.
#> 
#>          Estimate    SE
#> elpd_loo -14389.4  70.7
#> p_loo       157.8   2.6
#> looic     28778.9 141.3
#> ------
#> MCSE of elpd_loo is NA.
#> MCSE and ESS estimates assume independent draws (r_eff=1).
#> 
#> Pareto k diagnostic values:
#>                           Count Pct.    Min. ESS
#> (-Inf, 0.07]   (good)        0    0.0%  <NA>    
#>    (0.07, 1]   (bad)         0    0.0%  <NA>    
#>     (1, Inf)   (very bad) 9900  100.0%  <NA>    
#> See help('pareto-k-diagnostic') for details.
# }