R/plot.cmaRs.R

In cmaRs: Implementation of the Conic Multivariate Adaptive Regression Splines in R

#' A plot function designed for prediction of CMARS
#'
#' This function allows you to construct three different plots,
#' namely actual versus predicted response;
#' fitted values versus standardized residuals;
#' and standardized residuals versus order if the model is
#' constructed for prediction purpose. Moreover, Receiver
#' Operating Characteristic Curve of classification models
#' can be produced.
#' @param x A cmaRs object which is obtained by prediction.
#' @param ... Additional parameters.
#' @return An S3 model of class "plot.cmaRs"
#' @export
#' @examples
#' \dontrun{
#' # Without \code{MOSEK}, the example code is not executable.
#' # For installation of Mosek, plese see the documentation of 'Rmosek'.
#' data("trees", package = "datasets")
#' model.prediction <- cmaRs(Volume ~ ., degree = 5, nk = 20, data = trees)
#' plot.cmaRs(model.prediction)
#' }
#' @importFrom graphics par
#' @importFrom graphics abline
#' @importFrom graphics text
#' @importFrom ROCR prediction
#' @importFrom ROCR performance


plot.cmaRs <- function(x, ...) {
  if (x$classification == TRUE) {
    my.pred <- ROCR::prediction(x$fitted.values, as.factor(x$y))
    perf <- ROCR::performance(my.pred, "tpr", "fpr")
    plot(perf)
  } else {
    fitted.values <- x$fitted.values
    actual.values <- x$y
    r <- x$r
    residuals <- scale(x$residuals, scale = TRUE, center = TRUE)[, 1]

    oldpar <- graphics::par(no.readonly = TRUE)
    on.exit(graphics::par(oldpar))


    all.plots <- graphics::par(mfrow = c(3, 1))
    plot(actual.values, fitted.values,
      col = "firebrick", main = "Actual versus Predicted Response Values",
      xlab = "Actual Response", ylab = "Predicted Response"
    )
    graphics::abline(a = 0, b = 1, col = "gray0")
    graphics::text(
      x = min(actual.values) + 0.0001,
      y = max(fitted.values) - ((max(fitted.values) - min(fitted.values)) / 2),
      bquote(paste(rho(Y, hat(Y))) == .(r)), cex = 1, pos = 4
    )
    plot(fitted.values, residuals,
      col = "firebrick",
      main = "Fitted Values versus Standardized Residuals",
      xlab = "Fitted Values",
      ylab = "Standardized Residuals"
    )
    graphics::abline(h = 0, col = "black")
    plot(residuals,
      type = "o", col = "black",
      main = "Standardized Residuals versus Order",
      xlab = "Observation Order", ylab = "Standardized Residuals"
    )
    class(all.plots) <- "plot.cmaRs"
    invisible(all.plots)
  }
}