Develop a fuzzy expert system for detecting fraudulent claims for home insurance.

Assume the following seven input variables: the number of claims filed by a claimant in the last 12 months, amount of the current claim, how long a claimant was with the current insurer, average balance on all accounts over the last 12 months, the number of overdrafts over the last 12 months, annual income of a claimant, and whether a fundamental change occurred in the claimant’s life in the last four months (for instance, he or she got married or divorced, lost a job, or became a parent). Assume that the output of a fuzzy system is the likelihood of a fraud (a number between zero and one, with zero meaning that the insurance claim is non-fraudulent, and one that the likelihood of a fraud is very high).

Use a hierarchical fuzzy modelling approach. First, evaluate insurance history (use the first three input variables), banking history (use the fourth and fifth input variables), and claimant status (use the last two input variables). Then, evaluate fraud likelihood using insurance history, banking history and claimant status as inputs.