How to tune parameter kernels in SVM?

Question

I want with tune() from e1071 find optimal kernel from the list c('linear', 'polynomial', 'radial basis', 'sigmoid'). How to do it?

I tried like this, but it doesn’t work:

svmtune <- tune(svm, y~., data=dat, tunecontrol=tune.control(kernel=c('linear', 'polynomial',
                                                                      'radial basis', 'sigmoid')))

Error in tune.control(kernel = c("linear", "polynomial", "radial basis",  :
   unused argument (kernel = c("linear", "polynomial", "radial basis", "sigmoid"))

Answer 1

I found a way to solve this problem:

svmtune <- tune(svm, y~., data=dat, cost=10, ranges=list(kernel=c("linear", "polynomial",
                                                                  "radial", "sigmoid")))