library(e1071) svm_fit<-svm(y~x1+x2+x3,data=training) svm_predictions<-predict(svm_fit,newdata=testing) error<-sqrt((sum((testing$y-svm_predictions)^2))/nrow(testing)) error