R/tam_pv_mcmc_parameter_samples.R

In alexanderrobitzsch/TAM: Test Analysis Modules

## File Name: tam_pv_mcmc_parameter_samples.R
## File Version: 0.19

tam_pv_mcmc_parameter_samples <- function(beta_samples, variance_samples)
{
    require_namespace_msg("coda")
    #--- parameter_samples
    parameter_samples <- data.frame( beta_samples, variance_samples )
    saved_iter <- attr( beta_samples, "saved_iter")
    parameter_samples <- coda::mcmc( data=parameter_samples, start=min(saved_iter),
                            end=max(saved_iter), thin=1)

    #--- beta estimates
    beta_groups <- attr( beta_samples, "beta_groups")
    D <- attr( beta_samples, "D")
    G <- attr( beta_samples, "G")
    ncol_Y <- attr( beta_samples, "ncol_Y")
    beta_index <- attr( beta_samples, "beta_index")
    beta_M <- colMeans(beta_samples)
    beta <- list()
    if (beta_groups){
        for (gg in 1:G){
            ind_gg <- beta_index[[gg]]
            beta[[gg]] <- matrix( beta_M[ ind_gg ], nrow=ncol_Y, ncol=D )
        }
    }
    if ( ! beta_groups ){
        beta0 <- matrix( beta_M, nrow=ncol_Y, ncol=D )
        for (gg in 1:G){
            beta[[gg]] <- beta0
        }
    }

    #--- variance and correlation estimates
    correlation <- variance <- list()
    variance_index <- attr( variance_samples, "variance_index")
    variance_M <- colMeans(variance_samples)
    for (gg in 1:G){
        ind_gg <- variance_index[[gg]]
        variance_M_gg <- variance_M[ind_gg]
        var_gg <- tam_vec2symmmatrix(variance=variance_M_gg)
        variance[[gg]] <- var_gg
        cor_index <- tam_pv_mcmc_parameter_samples_correlation_index(index=ind_gg)
        cor_gg <- tam_pv_mcmc_parameter_samples_correlation(variance_samples=variance_samples,
                        cor_index=cor_index)
        correlation[[gg]] <- cor_gg
    }
    #--- OUTPUT
    res <- list(parameter_samples=parameter_samples, beta=beta, variance=variance,
                    correlation=correlation)
    return(res)
}