R/my_rf_cv2.R

In amakinney/Stat302Package: What the Package Does (One Line, Title Case)

#' Penguin body mass predictor
#'
#' This function predicts the penguins body mass based on physical trait inputs.
#'
#' @param k Numeric input representing the number of folds.
#'
#' @keywords prediction
#'
#' @return Numeric output representing the cv error.
#'
#' @examples
#'  my_rf_cv(5)
#'
#' @export
my_rf_cv <- function(k) {

  folds <- sample(rep(1:k, length = nrow(data_cleaned)))

  for(i in 1:30) {
    for (i in 1:k) {

    data_training <- data_cleaned %>% filter(folds != i)
    data_testing <- data_cleaned %>% filter(folds == i)

    modele <- randomForest(body_mass_g ~ bill_length_mm + bill_depth_mm +
                             flipper_length_mm, data = data_training, ntree = 100)

    predic <- predict(modele, data_testing[, -1])

    MSE <- (predic - data_cleaned$body_mass_g[i])^2
    }


  }


  avg_MSE <- sum(MSE) / k

  return(avg_MSE)

}