r - i do LOOCV in my data set ( svm classifier) and i need to extract the accuracy of prediction for Plot ROC curve

Question

i have a datasets with 587 rows and 0 column. i do LOOCV for random forest and using "a" vector for plot ROC curve. this is my code for random forest:

a=1:587
for (k in 1:587){
  set.seed(1) 
  a[k] <- predict(randomForest(death ~ ., 
                                         data = df2.imp$ximp[-k,], mtry=4),type="prob", 
                            newdata = df2.imp$ximp[k,,drop=F])[2] 
}

But this for loop doesn't work for SVM classifier . if i run this code , the results is null :

a=1:587
for (k in 1:587){
  set.seed(1) 
  a[k] <- predict(svm(death ~ ., 
                                         data = df2.imp$ximp[-k,]),type="prob", 
                            newdata = df2.imp$ximp[k,,drop=F])[2] 
}

it is important for me to do LOOCV and plot ROC curve . is there another way to do these instead?

Answer 1

等待大神答复