R/aldvmm.gof.R

In aldvmm: Adjusted Limited Dependent Variable Mixture Models

#' Calculating Goodness of Fit Measures
#'
#' \ifelse{html}{\code{\link[aldvmm]{aldvmm.gof}}}{\code{aldvmm.gof()}}
#' calculates residual- and likelihood-based goodness of fit measures.
#'
#' @param res a numeric vector of residuals of all observations in the
#'   estimation data.
#' @param ll a numeric value of the log-likelihood.
#' @param par a named numeric vector of parameter estimates.
#'
#' @details
#'   \ifelse{html}{\code{\link[aldvmm]{aldvmm.gof}}}{\code{aldvmm.gof()}}
#'   calculates mean squared errors as \eqn{MSE = \frac{\sum_{i = 1}^{n} (y_{i}
#'   - \hat{y}_{i})^2}{n - k}}{(sum(y - yhat)^2)/(n - k)}, and mean absolute
#'   errors as \eqn{MAE = \frac{\sum_{i = 1}^{n} y_{i} - \hat{y}_{i}}{n -
#'   k}}{sum(y - yhat)/(n - k)}, where \eqn{y_{i}}{y} denotes observed
#'   outcomes, \eqn{\hat{y}_{i}}{yhat} denotes fitted values, \eqn{n}{n}
#'   denotes the sample size, and \eqn{k}{k} denotes the number of parameters.
#'   The Akaike information criterion is calculated as \eqn{2k - 2ll}{2*k -
#'   2*ll}, and the Bayesian information criterion is calculated as
#'   \eqn{k\log(n) - 2ll}{k*log(n) - 2*ll}, where \eqn{ll}{ll} denotes the
#'   log-likelihood.
#'
#' @return \ifelse{html}{\code{\link[aldvmm]{aldvmm.gof}}}{\code{aldvmm.gof()}}
#'   returns a list including the following objects. \item{\code{mse}}{a
#'   numeric value of the mean squared error of observed versus fitted
#'   outcomes.} \item{\code{mae}}{a numeric value of the mean absolute error of
#'   observed versus fitted outcomes.} \item{\code{ll}}{a numeric value of the
#'   negative log-likelihood.} \item{\code{aic}}{a numeric value of the Akaike
#'   information criterion.} \item{\code{bic}}{a numeric value of the Bayesian
#'   information criterion.}
#'
#' @export

aldvmm.gof <- function(res,
                       par,
                       ll) {
  
  gof <- list()
  
  gof[["mse"]] <- sum(res^2) / (length(res) - length(par))
  gof[["mae"]] <- sum(abs(res)) / (length(res) - length(par))
  
  if (is.na(gof[['mse']])) {
    base::warning("no mse or mae were obtained\n", 
                  call. = FALSE)
  }
  
  gof[["ll"]] <- -ll
  gof[["aic"]] <- 2 * length(par) - 2 * ll
  gof[["bic"]] <- length(par) * log(length(res)) - 2 * ll
  
  return(gof)
  
}