9. Your company is launching a new service. Research indicates that if you target a specific demographic of your customers with mailers detailing the service, the rate of responses would increase. This is compared to the baseline of responses you will likely receive if you send out mailers without predictive models. You would like to know what percentage of the demographic should be targeted to yield the best results. Which predictive performance method can you use to determine the best outcome? 

A. Classification Matrix 

B. Lift Curve 

C. Mean Absolute Error 

D. ROC Curve 

E. Total Sum of Squared Errors 

10. In what circumstance is a business analyst most likely to accept a statistical model with an asymmetric misclassification rate? 

A. The specificity of the classifier is high; therefore, the analyst may be confident that records of lesser importance are rarely intermingled within the class of primary interest. 

B. The sensitivity of the classifier is high; therefore, the analyst may be confident that records of primary importance are rarely misclassified. 

C. Attempting to reduce the rate of error may have ethical implications. 

D. The cost associated with the misclassification of records with the high error rate is nominal. 

E. There’s no better model available; therefore, the risks of misclassification must be accepted as mandated by governmental regulations. 

11. Which of the following statements about data mining is FALSE? 

A. Intelligence agencies might use data mining to determine which of a huge quantity of intercepted communications are of interest. 

B. One can use logistic regression to identify which medical reimbursement applications have a higher probability of involving fraud.

 C. If applied correctly, all data mining methods lead to the same results. 

D. Linear regression can be used to predict a quantitative variable. 

E. Predictive analytics involves tasks such as classification and prediction that can produce business intelligence.