Consumer Credit Limit Assignment Using Bayesian Decision Theory and Fuzzy Logic – A Practical Approach

The market for consumer credit is diversified and multilayered. This makes it a challenging task to come up with a generalized theory of consumer credit limit assignment. However it is natural to surmise that an accurate and dynamic credit limit assignment strategy will certainly reduce the magnitude of the random and non-random uncertainty embedded in future credit limit management activities. Traditionally credit risk scorecards are employed for such strategic decisions. In order to minimize the misclassification error, Bayesian decision theory can be employed instead of scorecards for a more accurate credit limit strategy. However, estimation of the probability density function is a challenging task under practical circumstances. In this paper, we have proposed a modelling framework algorithm to estimate the class conditional density functions using frequency probability stemming from statistically independent simulations. Also for the continuous variable components within the feature vector, the class membership of a new entrant is being assigned using fuzzy logic. This makes the model robust, easy to handle, comprehend, implement and control