R/ef_error_handling.R

In UMeforensics/eforensics_public: Election Forensics: Positive Empirical Models of Election Fraud

ef_check_jags <- function()
{
    if (!runjags::testjags(silent=T)$JAGS.found) {
        stop("\n\nThe election forensics package current uses JAGS, but is was not found in the machine you are running the code. Please install it and run the code again.\n\n")
    }
    invisible()
}

check_mcmc <- function(mcmc)
{
    if (!"n.chains" %in% names(mcmc)) {stop("\n\nYou must include number of chains (n.chains) in the list provided in the mcmc parameter of the function eforensics(). It must be an interger equal or greater than 1. \n\n")}
    if (!"n.iter"   %in% names(mcmc)) {stop("\n\nYou must include number of iterations (n.iter) in the list provided in the mcmc parameter of the function eforensics() \n\n")}
    if (!"burn.in"  %in% names(mcmc)) {stop("\n\nYou must include the burn-in iterations (burn.in) in the list provided in the mcmc parameter of the function eforensics() \n\n")}
    if (!"n.adapt"  %in% names(mcmc)) {stop("\n\nYou must include number of adaptative steps (n.adapt) in the list provided in the mcmc parameter of the function eforensics() \n\n")}
}

check_nchains_for_psrf <- function(mcmc, mcmc.conv.diagnostic)
{
    if('n.chains' %in% names(mcmc)){
        if (!is.null(mcmc$n.chains)) {
            if (mcmc.conv.diagnostic=="PSRF" & mcmc$n.chains==1) {
                stop("\n\nYou need to run at least 2 chains to use PSRF as diagnostic. Set 'mcmc$n.chains' to be greater than 1 if mcmc.conv.diagnostic=='PSRF'\n\n")
            }
        }
    }
}

check_local_fraud <- function(samples)
{
    parameters.monitored = samples.par %>% colnames(.)  %>% stringr::str_replace(string=., pattern="\\[[0-9]*]", replacement="") %>% unique
    parameters.required  = ef_get_parameters_to_monitor(model) 
    if(!all(parameters.required %in% parameters.monitored))
        stop(paste0("\n\nYou must monitor all parameters to compute the distribuion of frauds at the observation level. Use parameters='all' in the function eforensics()\n\n") )
}