R/tam_pv_mcmc_postproc_ic.R

In TAM: Test Analysis Modules

## File Name: tam_pv_mcmc_postproc_ic.R
## File Version: 0.12

tam_pv_mcmc_postproc_ic <- function(parameter_samples, deviance_samples,
        theta_samples_mean, AXsi, B, guess, beta, variance, group_index, G, Y,
        resp, resp.ind, maxK, pv, resp_ind_bool )
{

    nstud <- nrow(Y)
    nitems <- ncol(resp)
    D <- attr( pv, "D")
    nplausible <- attr( pv, "nplausible")

    #--- init ic vector
    ic <- c()

    #*******************
    # inference based on marginal likelihood
    like <- tam_pv_mcmc_compute_marginal_likelihood( pv=pv, AXsi=AXsi, B=B, guess=guess,
                resp=resp, resp.ind=resp.ind, maxK=maxK, resp_ind_bool=resp_ind_bool )
    ic$deviance <- -2*sum( log( like ) )
    ic$n <- nstud
    ic$Npars <- ic$np <- ncol(parameter_samples)

    #-- compute all criteria
    ic <- tam_mml_ic_criteria(ic=ic)

    #*****************
    # fully Bayesian inference
    theta <- theta_samples_mean

    #--- Dbar
    ic$Dbar <- mean(deviance_samples)
    #--- Dhat
    like <- tam_pv_mcmc_evaluate_likelihood( theta=theta, AXsi=AXsi, B=B, guess=guess,
                resp=resp, resp.ind=resp.ind, maxK=maxK, resp_ind_bool=resp_ind_bool )
    ic$Dhat <- -2*sum( log(like) )
    #--- pD
    ic$pD <- ic$Dbar - ic$Dhat
    #--- DIC
    ic$DIC <- ic$Dhat + 2 * ic$pD
    #--- OUTPUT
    return(ic)
}

# z0 <- tamcat( label=" * rest", time0=z0, active=active)