1. A 2-NN (2-nearest neighbor) model is more likely to overfit than a 20-NN model. True or False?

2. Similarity measures can be incorporated into:

(A) Classification trees

(B) Logistic regressions

(C) Hierarchical clustering

(D) All of the above

3. Which of the following isNOTan issue when using k-Nearest Neighbor (k-NN)?

(A) Comparing instances may not be appropriate in all contexts

(B) It has trouble incorporating domain knowledge

(C) Too many irrelevant attributes can influence the results

(D) k-NN is difficult to compute and is not good when fast predictions are needed

4. Euclidean distance is the only feasible measure of similarity when usingk-NN techniques. True or False?

5. Which data mining technique is most appropriate for the following business question? "Of all my accounts, which are the most likely to exhibit fraud based on prior accounts that have and have not been defrauded?"

(A) Classification tree induction

(B) Hierarchical clustering

(C) Nearest neighbors

(D) Linear regression