case_bootstrap_constrained_lmm_em: Bootstrap for monotone-constrained polynomial LMM

In bonStats/gcreg: General Constraint REGression

Description

EXPIREMENTAL: Bootstrap for fitting polynomial constrained mixed effect model with monotonic mean and (potentially) monotonic random effects. For use with block-random effects (non-crossed), e.g. subject-specific random effects.

Usage

case_bootstrap_constrained_lmm_em(model, N, maxit = 25, tol = 0.01, start,
  verbose = F, save_steps = F, check_up = F, verbose_boot = verbose,
  parallel = F, cores = 1)

Arguments

model	model specification made by make_em_model_specs.
N	bootstrap size.
maxit	maximum number of iterations for each resample.
tol	relative tolerance for convergence of parameter values for each resample.
start	start values for parameters for each resample. Defaults to NULL.
verbose	output current iteration and quasi-log-likelihood value in each resample.
save_steps	should each em step be saved and return in a list?
check_up	should the algorithm check if the quasi-log-likelihood is improving each iteration?
verbose_boot	Output when a bootstrap resample is finished.
parallel	Execute in parallel. This will turn off verbose settings. ONLY IMPLEMENTED FOR LINUX/MAC
cores	Number of cores to use.

Details

The interface and default actions of this function are under development and may change without warning. See the mixed effects vignette for usage details. Uses case bootstrapping, where a sample of groups is chosen with replacement and the new estimates are used.

Value

list including estimated mixed effects model(s).

bonStats/gcreg documentation built on May 20, 2019, 5:44 p.m.