library(kernlab) ## Start input data load("iris.5f.RData") tra.iris.1 <- data.frame(iris.5.1.tra) tra.iris.2 <- data.frame(iris.5.2.tra) tra.iris.3 <- data.frame(iris.5.3.tra) tra.iris.4 <- data.frame(iris.5.4.tra) tra.iris.5 <- data.frame(iris.5.5.tra) names.attrs <- c("Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width", "Species") colnames(tra.iris.1) <- names.attrs colnames(tra.iris.2) <- names.attrs colnames(tra.iris.3) <- names.attrs colnames(tra.iris.4) <- names.attrs colnames(tra.iris.5) <- names.attrs tst.iris.1 <- data.frame(iris.5.1.tst[, 1 : 4]) tst.iris.2 <- data.frame(iris.5.2.tst[, 1 : 4]) tst.iris.3 <- data.frame(iris.5.3.tst[, 1 : 4]) tst.iris.4 <- data.frame(iris.5.4.tst[, 1 : 4]) tst.iris.5 <- data.frame(iris.5.5.tst[, 1 : 4]) names.outattr <- c("Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width") colnames(tst.iris.1) <- names.outattr colnames(tst.iris.2) <- names.outattr colnames(tst.iris.3) <- names.outattr colnames(tst.iris.4) <- names.outattr colnames(tst.iris.5) <- names.outattr real.iris.1 <- data.frame(matrix(iris.5.1.tst[, 5], ncol = 1)) real.iris.2 <- data.frame(matrix(iris.5.2.tst[, 5], ncol = 1)) real.iris.3 <- data.frame(matrix(iris.5.3.tst[, 5], ncol = 1)) real.iris.4 <- data.frame(matrix(iris.5.4.tst[, 5], ncol = 1)) real.iris.5 <- data.frame(matrix(iris.5.5.tst[, 5], ncol = 1)) ## generate model using classification SVM with default parameters (for simplicity). object.SVM.1 <- ksvm(Species ~., data = tra.iris.1, type = "C-bsvc", kernel = "rbfdot", kpar = list(sigma = 0.1), C= 10, prob.model = TRUE) object.SVM.2 <- ksvm(Species ~., data = tra.iris.2, type = "C-bsvc", kernel = "rbfdot", kpar = list(sigma = 0.1), C= 10, prob.model = TRUE) object.SVM.3 <- ksvm(Species ~., data = tra.iris.3, type = "C-bsvc", kernel = "rbfdot", kpar = list(sigma = 0.1), C= 10, prob.model = TRUE) object.SVM.4 <- ksvm(Species ~., data = tra.iris.4, type = "C-bsvc", kernel = "rbfdot", kpar = list(sigma = 0.1), C= 10, prob.model = TRUE) object.SVM.5 <- ksvm(Species ~., data = tra.iris.5, type = "C-bsvc", kernel = "rbfdot", kpar = list(sigma = 0.1), C= 10, prob.model = TRUE) pred.1 <- as.matrix(predict(object.SVM.1, tst.iris.1)) pred.2 <- as.matrix(predict(object.SVM.2, tst.iris.2)) pred.3 <- as.matrix(predict(object.SVM.3, tst.iris.3)) pred.4 <- as.matrix(predict(object.SVM.4, tst.iris.4)) pred.5 <- as.matrix(predict(object.SVM.5, tst.iris.5)) err.1 = 100*sum(real.iris.1!=pred.1)/nrow(real.iris.1) err.2 = 100*sum(real.iris.2!=pred.2)/nrow(real.iris.2) err.3 = 100*sum(real.iris.3!=pred.3)/nrow(real.iris.3) err.4 = 100*sum(real.iris.4!=pred.4)/nrow(real.iris.4) err.5 = 100*sum(real.iris.5!=pred.5)/nrow(real.iris.5) err.aver <- (err.1 + err.2 + err.3 + err.4 +err.5)/5 print(err.aver)