Using Rapidminer:

1. What is the optimal k?
2. Show the confusion matrix for the training+validation partition for the best k value obtained.
3. Report the performance metrics (accuracy, classification error, AUC, precision, recall, and f measure) for the training+validation partition for the best k value obtained.
4. Show the confusion matrix for the holdout partition.
5. Report the performance metrics (accuracy, classification error, AUC, precision, recall, and f measure) for the holdout partition.
6. Which model would you prefer, the optimal decision tree from Q2 or the optimal k-NN model? Why?