Exam MAS-I Practice Question 99
Study guide: Exam MAS-I
A homeowners claim-frequency GLM is fitted to five accident years of data. Over those years frequency trended upward, and the share of newer homes in the book grew steadily. Home age will be a rating variable; the rating plan will not vary by accident year. How should accident year be treated in the model?
- A) Include it as a categorical predictor, estimate its effect, and leave that effect out of the rates
- B) Leave it out, because a variable that will not be in the rating plan does not belong in the model
- C) Include it as an offset, so that its coefficient is fixed at 1 and nothing extra is estimated
- D) Include it as a categorical predictor and build its fitted relativities into the rates
- E) Use it as a prior weight, so that the most recent accident years count for more in the fit
Answer this question and see the worked solution