R/print.ModelSelectionEvaluation.R

In benmack/oneClass: One-class classification in the absence of test data

#' @name print.ModelSelectionEvaluation
#' 
#' @title Print an object of class \code{ModelSelectionEvaluation}
#' 
#' @description A table is printed with test and PU- performance metrics for the different evaluated models.
#' Test metrics: \cr
#' auc: auc
#' mxK.K: kappa at the threshold which leads to the maximum kappa
#' mxK.OA: Overall Accuracy at the threshold which leads to the maximum kappa
#' ...
#' 
#' @param x an object of class \code{ModelSelectionEvaluation}
#' @param by character giving a name of the table used to sort the table, default \code{NULL} is no sorting
#' @param digits number of digits to be printed
#' @param invisible do not print
#' @param ... options passed to the generic print method.
#' 
#' @method print ModelSelectionEvaluation
#' @export
print.ModelSelectionEvaluation <- function (x, by=NULL, digits=2, invisible=FALSE, ...) {
  evAtMaxK <- lapply(x$test, evaluateAtTh, th='max.kappa')
  mxK.K <- sapply(evAtMaxK, function(x) x$thDependent$kappa)
  mxK.OA <- sapply(evAtMaxK, function(x) x$thDependent$CCR)
  mxK.PA <- sapply(evAtMaxK, function(x) x$thDependent$TPR)
  mxK.UA <- sapply(evAtMaxK, function(x) x$thDependent$PPP)
  auc <- sapply(x$test, function(x) x@auc)
  
  df <- cbind(auc=auc, mxK.K=mxK.K, mxK.OA=mxK.OA, mxK.PA=mxK.PA, mxK.UA=mxK.UA, x$train)
  
  if (!is.null(by))
    df <- df[order(df[, by]), ]
  
  if (!invisible) {
    ans <- df
    num <- is.numeric(df[1, ])
    ans[, num] <- round(df[, num], 2)
    print(ans, ...)
  }
  
  invisible(df)
}