R/naive_model.R

In CorReg: Linear Regression Based on Linear Structure Between Variables

#' How would it be if we were naive ?
#' 
#' @description Describe the result of a naive binary discriminant model
#' 
#' @param proba_1 The ratio of 1 in the population (if nb_1 is NULL)
#' @param effectif The global effective of the population
#' @param nb_1 The number of 1 in the population. If not NULL proba_1 is not read.default=NULL
#' 
#' @export
naive_model <- function(proba_1 = 0.8, effectif = 100, nb_1 = NULL) {
  if (!is.null(nb_1)) {
    vrai = rep(0, times = effectif)
    vrai[1:nb_1] = 1
    proba_1 = nb_1 / effectif
  } else {
    vrai = runif(n = effectif, min = 0, max = 1)
    vrai[vrai > proba_1] = 0
    vrai[vrai != 0] = 1
  }
  pred = runif(n = effectif, min = 0, max = 1)
  pred[pred > proba_1] = 0
  pred[pred != 0] = 1
  count = table(pred, vrai)
  percent = count / effectif
  balance = data.frame(percent[1, 1] + percent[2, 2], percent[1, 2] + percent[2, 2])
  names(balance) = c("TRUE", "FALSE")
  
  return(list(count = count, percent = percent, balance = balance))
}