Exam MAS-I Practice Question 157
Study guide: Exam MAS-I
An insurer is choosing a classifier to flag claims for fraud review. For each description below, say which classifier it describes: naive Bayes (as in ISLR §4.4.4), quadratic discriminant analysis (QDA), both, or neither. An option may be used more than once or not at all.
Part a (1 point)
Estimates, separately for each class, the covariance between every pair of predictors.
- A) Naive Bayes only
- B) QDA only
- C) Both naive Bayes and QDA
- D) Neither
Part b (1 point)
Handles a qualitative predictor by using the share of each class's training observations at each of its levels.
- A) Naive Bayes only
- B) QDA only
- C) Both naive Bayes and QDA
- D) Neither
Part c (1 point)
Models P(Y = k \mid X = \mathbf{x}) directly as a function of \mathbf{x}, without specifying how the predictors are distributed within each class.
- A) Naive Bayes only
- B) QDA only
- C) Both naive Bayes and QDA
- D) Neither
Part d (1 point)
With normal predictor distributions whose variances differ by class, can produce a boundary between two classes that is curved in the predictors.
- A) Naive Bayes only
- B) QDA only
- C) Both naive Bayes and QDA
- D) Neither
Answer this question and see the worked solution