QuestionQ39

VerifiedAdvanced Analytics - Theory, Application, and Interpretation of Results for Eight Methods

Question Image

Refer to the exhibit.

To predict whether a customer will renew an annual property-insurance policy, an insurance company developed and operationalized a naïve Bayes classification model. The model assigns each customer one of two class labels, renewal or non-renewal, based on the customer’s attributes.

A subset of the key attributes, their values, and their associated conditional probabilities is shown in the exhibit. A customer has these attributes:

  • Age is greater than 65 years.
  • Owns their own home.
  • Renewal month is August.

If 20% of customers do not renew the policy each year, what is the score for renewal in the naïve Bayesian model for the described customer?

Explanation

The naïve Bayes renewal score is the renewal prior multiplied by the conditional probability of each observed attribute under the renewal class: (P(\mathrm{Renewal}) \times P(>65\mid\mathrm{Renewal}) \times P(\mathrm{Own}\mid\mathrm{Renewal}) \times P(\mathrm{August}\mid\mathrm{Renewal})). With (P(\mathrm{Renewal})=0.8), the score is (0.8 \times 0.3 \times 0.9 \times 0.1=0.0216).

Community Discussion

No comments yet. Be the first to start the discussion!