Check model identifiability — tobs_check

Diagnostics for potential identifiability issues. Checks for confounded covariates, low detection or detection-free sites, and sparse replication.

Usage

tobs_check_id(model, fit = NULL)

Arguments

model: A tobs_model object (before fitting).
fit: Optional tobs_fit object (for post-fit diagnostics).

Value

A list with identifiable, the issues messages, and prefit_checked (FALSE where the family carries no site-by-visit grid).

Details

Applies to the families whose response is a site-by-visit grid: occu(), abun(), removal() and fp_occu(). On any other family the pre-fit checks are reported as not applicable rather than silently skipped – an empty issues on a family the checks never ran on is not a clean bill of health. The post-fit sampler checks apply to any NUTS fit.

Examples

# \donttest{
sim <- simulate_occu(N = 100, J = 3, n_occ_covs = 1, n_det_covs = 1,
                     seed = 1)
fit <- tobs(~ occ_cov1, data = sim$data, family = occu(),
            detection = ~ det_cov1, y = sim$y, method = "laplace",
            control = list(verbose = FALSE, progress = FALSE))
chk <- tobs_check_id(fit$model, fit)
chk$identifiable
# }