R/confusion_matrix.R

In fairmodels: Flexible Tool for Bias Detection, Visualization, and Mitigation

#' Confusion matrix
#'
#' Calculates confusion matrix for given cutoff
#'
#' @param probs numeric, vector with probabilities given by model
#' @param observed numeric, vector with actual values from outcome, either 0 or 1
#' @param cutoff numeric, single value denoting cutoff/threshold
#'
#' @return object of class \code{confussion_matrix}
#' It is a list with following fields:
#' \itemize{
#' \item{tp}{number of True Positives}
#' \item{fp}{number of False Positives}
#' \item{tn}{number of True Negatives}
#' \item{fn}{number of False Negatives}
#' }
#'
#' @export
#' @rdname confusion_matrx
#'
#'
#' @examples
#'
#' probs <- rnorm(20, 0.4, 0.1)
#' observed <- round(runif(20))
#'
#' confusion_matrix(probs, observed, 0.5)
confusion_matrix <- function(probs, observed, cutoff) {
  stopifnot(length(probs) == length(observed))
  stopifnot(is.numeric(probs) & is.numeric(observed))
  stopifnot(is.numeric(cutoff))

  tp <- sum((observed == 1) * (probs >= cutoff))
  fp <- sum((observed == 0) * (probs >= cutoff))
  tn <- sum((observed == 0) * (probs < cutoff))
  fn <- sum((observed == 1) * (probs < cutoff))

  confusion_matrix <- list(tp = tp, fp = fp, tn = tn, fn = fn)
  class(confusion_matrix) <- "confusion_matrix"
  return(confusion_matrix)
}