Skip to contents

Compare models by information criteria

Source: R/diagnostics_generic.R
compare_models.Rd

Rank fitted models best-first by an information criterion. "waic" and "loo" use the native pointwise log-likelihood layer (tulpa_criteria() / tulpa_psis()) – WAIC and PSIS-LOO respectively, computed from each fit's [n_draws x n_obs] pointwise log-likelihood (fit$draws$log_lik), with no loo package dependency. "loglik" returns the (integrated) joint log-likelihood with the parameter count. A fit carrying no pointwise log-likelihood (a deterministic / point approximation) yields NA criterion columns rather than an error, so the table always has one row per model.

Usage

compare_models(..., criterion = c("waic", "loo", "loglik"))

Arguments

...

Named tulpa_fit objects.

criterion

"waic" (default), "loo", or "loglik".

Value

A data frame. For "loglik": model, n_params, logLik. For "waic" / "loo" (ranked best-first): model, elpd, se_elpd, p_eff, ic (-2 * elpd), delta (elpd gap to the best model), se_diff (SE of that pointwise elpd difference), and weight (the Akaike-style weight on the criterion).

See also

model_average() for model-averaged predictions, tulpa_criteria() and tulpa_psis() for the native criteria layer.

Examples

# \donttest{
set.seed(1)
df <- data.frame(x = rnorm(120))
df$y <- rpois(120, exp(0.4 + 0.5 * df$x))
f1 <- tulpa(y ~ x, data = df, family = "poisson")
f2 <- tulpa(y ~ 1, data = df, family = "poisson")
compare_models(full = f1, null = f2, criterion = "waic")
# }