Skip to contents

Parametric goodness-of-fit from simulated datasets

Source: R/goodness_of_fit.R
goodness_of_fit.Rd

Simulate datasets from a fitted model, refit the model on each simulated dataset, and compare simulated log-likelihood values to the original fit.

Usage

goodness_of_fit(object, nsim, seed = NULL, fit_args = list())

Arguments

object

Output of lifelihood().

nsim

Number of simulated datasets to generate and refit.

seed

Optional integer seed for reproducibility.

fit_args

Named list of additional arguments passed to lifelihood() when refitting each simulated dataset. By default, refits use n_fit = 1, MCMC = 0, and se.fit = FALSE for speed.

keep_fits

Whether to store fitted objects for each successful simulation. Default is FALSE.

Value

A lifelihoodGOF object (list) with:

  • original_loglik: log-likelihood of the original fit

  • simulated_loglik: numeric vector of simulated/refitted log-likelihoods

  • n_success: number of successful refits

  • n_failed: number of failed refits

  • p_lower_or_equal: proportion of simulated log-likelihoods lower than or equal to the original value

  • errors: per-simulation error messages (if any)

  • fits: optional list of fitted objects (only if keep_fits = TRUE)