R/AllMetrics.R

In AllMetrics: Calculating Multiple Performance Metrics of a Prediction Model

#'@title Calculating Multiple Performance Metrics of a Prediction Model
#' @description This provides a function to calculate multiple performance metrics for actual and predicted values.
#' @param actual This is the actual time series values
#' @param predicted This is the predicted values of a time series using a model
#'
#' @return
#' \itemize{
#'   \item AllMetrics - A data frame containing two columns with first column as the name of the eight metrics and second column as the corresponding values
#' }
#' @export
#'
#' @examples
#' actual <- c(1.5, 2.3, 25, 52, 14)
#' predicted <- c(1.2, 10, 3.5, 4.3, 5.6)
#' # Inside the function 1st specify actual then predicted
#' print(all_metrics(actual, predicted))
#' @references
#' \itemize{
#'\item Garai, S., & Paul, R. K. (2023). Development of MCS based-ensemble models using CEEMDAN decomposition and machine intelligence. Intelligent Systems with Applications, 18, 200202.
#'\item Garai, S., Paul, R. K., Kumar, M., & Choudhury, A. (2023). Intra-annual National Statistical Accounts Based on Machine Learning Algorithm. Journal of Data Science and Intelligent Systems. https://doi.org/10.47852/bonviewJDSIS3202870
#'\item Garai, S., Paul, R.K., Yeasin, M., Paul, A.K. (2024). CEEMDAN-Based Hybrid Machine Learning Models for Time Series Forecasting Using MARS Algorithm and PSO-Optimization. Neural Processing Letters, 56, 92. https://doi.org/10.1007/s11063-024-11552-w
#' }
# all metrics
all_metrics <- function(actual, predicted) {
  residuals <- actual - predicted
  abs_residuals <- abs(actual - predicted)

  # Replace actual with 1e-10 where actual is exactly zero
  actual[actual == 0] <- 1e-10

  scaled_abs_residuals <- abs_residuals / actual
  lag_frame <- data.frame(embed(actual, 2))
  diff <- lag_frame[, 1] - lag_frame[, 2]
  abs_diff <- abs(diff)
  mse <- mean(residuals^2)
  rmse <- sqrt(mse)
  rrmse <- 100 * rmse / mean(actual)
  mae <- mean(abs_residuals)
  mape <- 100 * mean(scaled_abs_residuals)
  mase <- mae / mean(abs_diff)
  nse <- 1 - (mse / (mean(actual^2) - (mean(actual))^2))
  wi <- 1 - (mse / mean((abs(actual - mean(actual)) + abs(predicted - mean(actual)))^2))
  lme <- 1 - mae / mean(abs(actual - mean(actual)))

  AllMetrics <- data.frame(cbind(c("RMSE", "RRMSE", "MAE", "MAPE", "MASE", "NSE", "WI", "LME"),
                                 c(round(rmse, 3), round(rrmse, 3), round(mae, 3), round(mape, 3),
                                   round(mase, 3), round(nse, 3), round(wi, 3), round(lme, 3))))
  colnames(AllMetrics) <- c("Metrics", "Values")
  dimnames(AllMetrics)
  dim(AllMetrics)
  return(AllMetrics)
}