1. The "hit rate" of a model is the ratio of:

A.) True positive to total positive instances

B.) True positive to false negative instances

C.) False positive to false negative instances

D.) True positive to false positive instances

2. An area under the ROC curve (AUC) of 1 means that is a classifier is no better than picking randomly. True or False?

3. A classifier with a hit rate of 60% when targeting 20% of the population of interest would have a lift of:

A.) 1/3

B.) 3

C.) 80

D.) 1,200

4. An online retailer is interested in sending e-coupons to existing customers likely to use the coupon. The product normally retails for $25 but can be purchased for $20 with the e-coupon. It costs $5 to make the product and nothing to email the e-coupon. If a cost/benefit matrix were created for this situation, what is the value of a false negative outcome?

A.) $20

B.) $15

C.) $5

D.) $0

5. Confusion matrices and cost/benefit matrices can be used to find the expected profit from sending a targeted advertisement. True or False?