Design a Bayesian network bank loan credit system which takes as input details about the customer (banking history, income, years in job, etc.) and details of a proposed loan to the customer (type of loan, interest rate, payment amount per month, etc.)

and estimates the probability of a loan default over the life of a loan. In order to parameterize this network you will probably have to make guestimates based upon aggregate statistics about default rates for different segments of your community, since banks are unlikely to release their proprietary data about such things.