R/calculateMAEs.R

In svcuonghvktqs/FORA: A package for time sereis forecast accuracy evaluation

#' Mean Absolute errors for the given data from dataset(MAEs)
#'
#' This function calculates and returns list of two dataframes,
#' where the first data frame contains MAEs for the given data, diferent horizons and methods,
#' the second one contains ranked list of the methods according to MAEs.
#' Also the function plots MAEs for different hirizons and methods.
#'
#' @aliases calculateMAEs
#' @param frame A data frame containing columns "actual", "forecast", "method", and "horizon".
#' @param sort logical. If TRUE the resulting list of MAEs dataframe and ranked dataframe of MAEs sorting by average value.
#' @return \code{calculateMAEs} function calculates and returns list of two dataframes,
#' where the first data frame contains MAEs for the given data, diferent horizons and methods,
#' the second one contains ranked dataframe of the methods according to MAEs.
#' Also the function plots MAEs for different hirizons and methods.
#' @author Sai Van Cuong, Maixm Shcherbakov and Andrey Davydenko
#' @seealso \code{\link{calculateAvgRelMAEs}}, \code{\link{calculateGMAPEs}}, \code{\link{calculateGMRAEs}}, \code{\link{calculateMAD_MEAN_ratio}},
#'  \code{\link{calculateMAPEs}}, \code{\link{calculateMASEs}}, \code{\link{calculateMdAPEs}}, \code{\link{calculateMPEs}},
#'  \code{\link{calculateMSEs}}, \code{\link{calculatePB_MAEs}}, \code{\link{calculateRMSEs}}, \code{\link{calculateSMAPEs}},
#'  \code{\link{calculateSMdAPEs}}.
#' @references Rob J. Hyndman, Anne B. Koehler (2006) Volume title: "International Journal of Forecasting".
#' Chapter title: \emph{Another look at measures of forecast accuracy}. Chapter pages : (p.679-688).
#' \url{http://eva.fcea.edu.uy/pluginfile.php/109034/mod_resource/content/0/2006_Hyndman_Predicc.pdf}.
#' @keywords dataframe
#' @examples
#' calculateMAEs(frame = FORAYearForecast)
#' calculateMAEs(frame = FORAYearForecast, sort = TRUE)
#' data1 <- subset(FORAYearForecast, actual >= 5000| forecast < 8000)
#' data2 <- FORAYearForecast[1:300,]
#' calculateMAEs(frame = data1, sort = TRUE)
#' calculateMAEs(frame = data2, sort = TRUE)
#'
#' @export
calculateMAEs <- function(frame, sort = FALSE){
  out <-matrix(NA, nrow = length(unique(frame$method)), ncol = length(unique(frame$horizon)))
  methodlist <- list()
  horizonlist <- list()
  MAElist <- list()
  MAE <- c()
  df = data.frame(out)
  colnames(df) <- paste("horizon = ", 1:length(unique(frame$horizon)), sep ="")
  rownames(df) <- unique(frame$method)
  ranks = data.frame(out)
  colnames(ranks) <- paste("horizon = ", 1:length(unique(frame$horizon)), sep ="")
  rownames(ranks) <- unique(frame$method)
  outlist <- list()

  for(j in as.vector(unique(frame$horizon))){
    for(i in as.vector(unique(frame$method))){
      df[i, j] <-  mean(abs((subset(frame, method == i & horizon == j)$actual - subset(frame, method == i & horizon == j)$forecast)), na.rm=TRUE)
    }
  }

  for (k in 1:length(unique(frame$horizon))){
    ranks[,k] <- rank(df[, k])
  }
  averagerank <- rowMeans(ranks, na.rm =TRUE)
  averageMAE <- rowMeans(df, na.rm =TRUE)
  ranks <- cbind(ranks, "average rank" = averagerank)
  df <- cbind(df, " average MAE" = averageMAE)

  for(m in 1:length(unique(frame$method))){
    MAElist[[m]] <- unname(df[m, 1:length(unique(frame$horizon))])
    methodlist[[m]] <- rep(as.vector(unique(frame$method))[m],length(unique(frame$horizon)))
    horizonlist[[m]]<- as.vector(unique(frame$horizon))

  }
  MAE1 <- Reduce(c, MAElist)
  MAE <- Reduce(c, MAE1)
  horizon <- Reduce(c, horizonlist)
  method = Reduce(c, methodlist)
  df2 <- data.frame(MAE, horizon, method )
# plots MAEs frame
  gp <-ggplot2::ggplot(df2, ggplot2::aes(x=horizon, y=MAE, group=method,color=method, shape=method)) +
    ggplot2::scale_shape_manual(values=1:nlevels(df2$method)) +
    ggplot2::labs(title = "MAE for different horizons and methods") +
    ggplot2::geom_line() +
    ggplot2::geom_point(size=3)+
    ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5))
  print(gp)
  outlist <- list("MAE" = df,"rank" =ranks)
  if(sort == FALSE){
    return(outlist)
  }else{
    frame1 <-df[order(df$` average MAE`),]
    frame2 <- ranks[order(ranks$`average rank`),]
    outlist <- list("MAE" = frame1,"rank" = frame2)
    return(outlist)
  }
}