Parametric bootstrap inference for fitted models — bootstrap

bootstrap_inference() simulates responses from a fitted gamlss.longitudinal model, refits the same model to each simulated response, and summarizes the bootstrap distribution of selected fixed coefficients. It is intended for opt-in applied uncertainty checks and should be run with enough replicates outside CRAN-time tests for final reporting.

Usage

bootstrap_inference(
  object,
  R = 100,
  terms = NULL,
  level = 0.95,
  seed = NULL,
  fit_args = list(),
  keep_fits = FALSE,
  ...
)

Arguments

object: A fitted gamlss.longitudinal object.
R: Number of bootstrap replicates.
terms: Optional coefficient names, formula-term names such as "mu.treatment", coefficient-name prefixes, or numeric indices to summarize.
level: Confidence level for percentile intervals.
seed: Optional random seed.
fit_args: Optional named list of arguments passed to each refit, such as max_outer_iter, max_inner_iter, or convergence tolerances.
keep_fits: Logical; keep successful refitted model objects.
...: Additional arguments passed to simulate.gamlss.longitudinal().

Value

An object of class gamlss_longitudinal_bootstrap.