| > #Find the best parameter gamma&cost |
| > p<-seq(-1,1,1) |
| > obj<-tune.svm(y~., data=train, sampling="cross", gamma=2∧(p), cost=2∧(p)) |
| > obj |
| Parameter tuning of ‘svm’: |
| - sampling method: 10-fold cross validation |
| - best parameters: |
| gamma cost |
| 0.5 2 |
| > #Building the SVM model |
| > svm.model<-svm(y~., data=train, type="C-classification", gamma=obj$best.parameters1, cost=obj$best.parameters2) |
| > #Classification capability: Train |
| > svm.pred<-predict(svm.model, train) |
| > tab<-table(predict=svm.pred, true=train,1) |
| > tab |
| true |
| predict 0 1 |
| 0 17 0 |
| 1 0 31 |
| > cat(Accurate Classification Rate = ,100sum(dig(tab))/sum(tab), % n) |
| Accurate Classification Rate = 100 % |
| > #Classification capability: Test |
| > svm.pred<-predict(svm.model, test) |
| > tab<-table(predict=svm.pred, true=test,1) |
| > tab |
| true |
| predict 0 1 |
| 0 2 1 |
| 1 6 15 |
| > cat(Accurate Classification Rate = ,100sum(dig(tab))/sum(tab), % n) |
| Accurate Classification Rate = 70.83333 % |
