A00-240 SAS Statistical Business Analysis Using SAS 9: Regression and Modeling

Two competing models are compared using ROC curves. Candidate 2 has a greater area under the curve than Candidate 1. Which model is the champion, and why?

• A. Candidate 1, because the area outside its curve is greater
• B. Candidate 2, because its area under the curve is greater
• C. Candidate 1, because it is closer to the diagonal reference line
• D. Candidate 2, because it necessarily shows less overfitting

A marketing campaign mails brochures for an expensive product. Mailing and production cost $50 per customer, and each sale generates $500 in revenue. Which profit matrix is correct?

• A. No mail: no purchase −$50, purchase $0; Mail: no purchase $0, purchase $450
• B. No mail: no purchase $0, purchase $0; Mail: no purchase −$50, purchase $500
• C. No mail: no purchase $0, purchase $0; Mail: no purchase −$50, purchase $450
• D. No mail: no purchase −$50, purchase $0; Mail: no purchase $0, purchase $500

A confusion matrix is created after oversampling a rare target class. Which pair of measures is not affected by this oversampling?

• A. Sensitivity and positive predictive value (PV+)
• B. Specificity and negative predictive value (PV−)
• C. Positive predictive value (PV+) and negative predictive value (PV−)
• D. Sensitivity and specificity

A logistic model was fit on data with 40% events, but the deployment population has only 5% events. The profit margin for correctly targeting an event is nine times the loss from incorrectly targeting a non-event. In the SAS scoring program, which values should be used for X and Y?

• A. X=40, Y=10
• B. X=.05, Y=10
• C. X=.05, Y=.40
• D. X=.10, Y=.05

The modeling data are split into training, validation, and test sets. Which set is best for model assessment?

• A. Training data
• B. All data combined
• C. Test data
• D. Validation data