library(randomForest) library(frbs) ## Input data data(frbsData) data.train <- frbsData$MackeyGlass1000.dt[1: 500, ] data.fit <- data.train[, 1 : 4] data.tst <- frbsData$MackeyGlass1000.dt[501 : 1000, 1 : 4] real.val <- matrix(frbsData$MackeyGlass1000.dt[501 : 1000, 5], ncol = 1) range.data<-matrix(c(0.43462, 1.3105, 0.43462, 1.3105, 0.43462, 1.3105, 0.43462, 1.3105, 0.43462, 1.3105), nrow=2) colnames(data.train) <- c("d1", "d2", "d3", "d4", "d5") colnames(data.fit) <- c("d1", "d2", "d3", "d4") colnames(data.tst) <- c("d1", "d2", "d3", "d4") ## generate model using classification tree with default parameters (for simplicity). object.rF <- randomForest(d5 ~., data.train, importance =TRUE, proximity = TRUE) pred <- predict(object.rF, data.tst) benchmark <- cbind(real.val, pred) colnames(benchmark) <- c("real", "tree") print(benchmark) residuals <- (real.val - pred) MSE <- mean(residuals^2) RMSE <- sqrt(mean(residuals^2)) SMAPE <- mean(abs(residuals)/(abs(real.val) + abs(pred))/2)*100 err <- c(MSE, RMSE, SMAPE) names(err) <- c("MSE", "RMSE", "SMAPE") print("Error Measurement: ") print(err)