R/plotParameters.R

In TreeBUGS: Hierarchical Multinomial Processing Tree Modeling

############## PLOTTING FUNCTIONS FOR betaMPT

#' Plot Parameter Estimates
#'
#' Plot parameter estimates for hierarchical MPT models.
#'
#' @param x a fitted Beta or latent-trait MPT model
#' @param includeIndividual whether to plot individual estimates
#' @param addLines whether to connect individual parameter estimates by lines
#' @param estimate type of point estimates for group-level and individual parameters
#'     (either \code{"mean"} or \code{"median"})
#' @param select character vector of parameters to be plotted (e.g., \code{select = c("d", "g")}. Can be used to plot subsets of parameters and change the order of parameters.
#' @param ... further arguments passed to the standard \code{\link{plot}} function
#'
#' @author Daniel Heck
#' @seealso \code{\link{betaMPT}}, \code{\link{traitMPT}}, \code{\link{plotDistribution}}
#' @examples
#' \dontrun{
#' plotParam(fit,
#'   addLines = TRUE,
#'   estimate = "median",
#'   select = c("d1", "d2")
#' )
#' }
#' @export
plotParam <- function(x, includeIndividual = TRUE, addLines = FALSE,
                      estimate = "mean", select = "all", ...) {
  stat <- ifelse(estimate == "median", "50%", "Mean")
  par.group <- x$summary$groupParameters$mean
  par.ind <- x$summary$individParameters
  parnames <- substr(rownames(par.group), 6, 100)
  if (select[1] == "all") {
    select <- parnames
  } else {
    if (!all(select %in% parnames)) {
      stop(
        "Check arguments: Not all parameters in 'select' are included in the MPT model!\n",
        "Parameters are: ", paste(parnames, collapse = ", ")
      )
    }
    par.group <- par.group[paste0("mean_", select), , drop = FALSE]
  }
  dims <- dim(par.ind)
  S <- nrow(par.group) # parameters
  N <- dims[2] # persons
  means <- par.group[, stat]

  plot(1:S, means,
    ylim = 0:1, xlim = c(.5, S + .5), pch = 19, xaxt = "n", # size=3,
    xlab = "MPT Parameters",
    ylab = paste0("Estimate (", estimate, "s)"), col = 2,
    main = paste0(
      "Group-level ", estimate, "s + 95% CI (red)",
      ifelse(includeIndividual,
        paste0(" and individual ", estimate, "s (gray)"),
        ""
      )
    ), ...
  )
  axis(side = 1, at = 1:S, labels = select)
  if (includeIndividual) {
    for (i in 1:N) {
      if (addLines) {
        lines(1:S + .05, par.ind[select, i, stat],
          col = adjustcolor(col = "black", alpha.f = .5)
        )
        points(1:S + .05, par.ind[select, i, stat],
          cex = .9, pch = 16,
          col = adjustcolor(col = "black", alpha.f = .5)
        )
      } else {
        points(1:S + seq(-.2, .2, length.out = N)[i],
          col = adjustcolor(col = "black", alpha.f = .5), # col=rainbow(N, alpha=.4)[i],
          pch = 16,
          par.ind[select, i, stat], cex = .9
        )
      }
    }
    points(1:S, means, cex = 1.3, col = 2, pch = 19)
  }
  segments(
    x0 = 1:S, y0 = par.group[, 3],
    y1 = par.group[, 5], lwd = 2, col = 2
  )
}