Run multistage GSI model

Usage

msgsi_mdl(
  dat_in,
  nreps,
  nburn,
  thin,
  nchains,
  nadapt = 0,
  keep_burn = FALSE,
  cond_gsi = TRUE,
  file_path = NULL,
  seed = NULL,
  iden_output = TRUE,
  p1_prior_weight = NULL,
  p2_prior_weight = NULL
)

Arguments

dat_in: Name of the input data.
nreps: Total number of iterations (includes burn-ins).
nburn: Number of warm-up runs.
thin: Frequency to thin the output.
nchains: Number of independent MCMC processes.
nadapt: Number of adaptation run (default is 0). Only available when running model in fully Bayesian mode.
keep_burn: To save the burn-ins or not (default is FALSE).
cond_gsi: To run the model in conditional GSI mode (default is TRUE).
file_path: File path to save the output. Leave it empty is you don't want to save the output.
seed: Random seed for reproducibility. Default is NULL (no random seed).
iden_output: Option to have trace history for individual assignments included in the final output. Default is TRUE.
p1_prior_weight: An optional tibble to specify weight for each broad-scale reporting group. Columns are repunit, grpvec, and weight.
p2_prior_weight: An optional tibble to specify weight for each regional reporting group. Columns are repunit, grpvec, and weight.

Value

A list contains reporting group proportion summary and trace for tier 1 (summ_t1, trace_t1), tier 2 (summ_t2, trace_t2) and two tiers combined (summ_comb, trace_comb), and record of individual assignment during first tier for each individual (idens).

Examples

# setup input data
msgsi_dat <-
  prep_msgsi_data(mixture_data = mix,
  baseline1_data = base_templin, baseline2_data = base_yukon,
  pop1_info = templin_pops211, pop2_info = yukon_pops50, sub_group = 3:5)
#> Compiling input data, may take a minute or two...
#> Time difference of 9.075418 secs

# run multistage model
msgsi_out <- msgsi_mdl(msgsi_dat, nreps = 25, nburn = 15, thin = 1, nchains = 1)
#> Running model (and the category is... Face!)
#> Time difference of 1.447003 secs
#> April-25-2025 19:09