Skip to contents

Run CD4 back-calculation model

Source: R/run_backcalc.R
run_backcalc.Rd

This function runs the CD4 back-calculation model using Stan and runs the post-processing.

Usage

run_backcalc(
  stan_data,
  migration = NULL,
  rita = NULL,
  age = NULL,
  inf_model = 1,
  diag_model = 1,
  model_seed = NULL,
  iter_warmup = 1000,
  iter_sampling = 1000,
  chains = 4,
  adapt_delta = 0.95,
  max_treedepth = 12,
  refresh = 100,
  messages = TRUE,
  start_yr = 0,
  nworkers = 1,
  chain_results = FALSE,
  return_fit = TRUE,
  checkpoint_dir = NULL,
  iter_per_chunk = NULL,
  init = NULL
)

Arguments

stan_data

A list of data formatted for Stan.

migration

Logical indicating whether to include migration in the model.

rita

Logical indicating whether to include RITA data in the model.

age

Logical indicating whether to include age-specific data in the model.

inf_model

Infection model type (1 = spline, 2 = random walk, 3 = GP).

diag_model

Diagnosis model type (1 = spline, 2 = random walk).

model_seed

Seed for model fitting (will be set randomly if NULL).

iter_warmup

Number of warmup iterations for Stan sampling.

iter_sampling

Number of sampling iterations for Stan sampling.

chains

Number of chains for Stan sampling.

adapt_delta

Adapt delta parameter for Stan sampling (default is 0.95).

max_treedepth

Max tree depth for Stan sampling (default is 12).

refresh

Number of iterations between progress messages.

messages

Logical indicating whether to show messages during sampling.

start_yr

Starting year for posterior predictive quantities.

nworkers

Number of parallel workers to use.

chain_results

Logical indicating whether to return results by chain.

return_fit

Logical indicating whether to return the fit object

checkpoint_dir

Directory for saving/loading checkpoints. If NULL (default), no checkpointing is performed.

iter_per_chunk

Number of sampling iterations per chunk when checkpointing. If NULL, all sampling iterations are run in a single chunk.

init

Initial values passed to Stan. When checkpointing, these are only used when starting a new run; resumed checkpointed fits use the saved checkpoint state.

Value

List of posterior predictive quantities and (optionally) the stanfit object.

Examples

sim_diags <- simulate_diagnoses(sim_seed = 123)
hiv_list <- run_backcalc(
 sim_diags, iter_warmup = 10, iter_sampling = 10, chains = 1, messages = FALSE, nworkers = 1
)
#> CmdStan model cache dir: /tmp/RtmpJloCEK/cd4backcalc-cmdstan
#> R tempdir(): /tmp/RtmpJloCEK
#> Error: CmdStan path has not been set yet. See ?set_cmdstan_path.