R/my_knn_cv.R

In alishaluo/STAT302package: Package Building for STAT 302

#' k-Nearest Neighbors Cross Validation Function
#'
#' This function uses k-nearest neighbors cross validation to predict the output
#' class.
#'
#' @param train data frame input.
#' @param  cl true class value of the training data.
#' @param k_nn numeric input of the number of neighbors.
#' @param k_cv numeric input of the number of folds.
#' @keywords prediction
#'
#' @return A list with the predicted class, and a numeric output of the cross
#'   validation error.
#'
#' @examples
#' my_knn_cv(train = my_gapminder[, c(4, 6)], cl = my_gapminder$continent,
#'           k_nn = 5, k_cv = 5)
#'
#' @import class gapminder magrittr stats dplyr
#' @export
my_knn_cv <- function(train, cl, k_nn, k_cv) {
  my_gapminder <- my_gapminder
  n <- nrow(my_gapminder)
  # randomly selects folds
  folds <- sample(rep(1:k_cv, length = n))
  # split data
  data <- data.frame(train, "split" = folds)
  miss_rate <- rep(NA, k_cv)
  train_err <- rep(NA, k_cv)
  for(i in 1:k_cv) {
    # X_i
    data_train <- data %>% dplyr::filter(split != i)
    # X_i^*
    data_test <- data %>% dplyr::filter(split == i)
    # Y_i
    cl_train <- cl[folds != i]
    # Y_i^*
    cl_test <- cl[folds == i]
    # remove split columns
    data_train$split <- NULL
    data_test$split <- NULL
    # predicts output class, Y_hat^*
    knn_output <- knn(train = data_train, test = data_test, cl = cl_train,
                      k = k_nn)
    # stores the proportion observations classified incorrectly
    miss_rate[i] <- mean(knn_output != cl_test)
   }
  # stores output for full data
  class <- knn(train = train, cl = cl, test = train, k = k_nn)
  # calculates average of the misclassication rate
  cv_err <- mean(miss_rate)
  # stores results into a list
  result <- list("class" = class, "CV_error" = cv_err)
  # returns list
  return(result)
}