svmtest0.R

library(e1071)
library(rpart)
data(Glass,package="mlbench")

##split data into a train and test set
index <- 1:nrow(Glass)
testindex <- sample(index, trunc(length(index)/3))
testset <- Glass[testindex,]
trainset <- Glass[-testindex,]
#print(trainset)

##svm
svm.model <- svm(Type ~ .,data = trainset, cost = 100, gamma = 1)
svm.pred <- predict(svm.model, testset[,-10])

##rpart
rpart.model <- rpart(Type ~ .,data = trainset)
rpart.pred <- predict(rpart.model, testset[,-10],typr = "class")

##compute svm confusion martrix
table(pred = svm.pred, true = testset[,10])

##compute rpart confusion martrix
#testset[,-10]=t(testset[,-6])
table(pred = rpart.pred, true = testset[,10])