Use of Soft Computing Techniques in Credit Risk Management of an Indian Bank: Application of Artificial Neural network

Abstract

This study uses ten variables from an Indian commercial bank in India for its retail customers. The study uses a Multilayer Perceptron (MLP) - Artificial Neural network architecture to understand the usability of such data for credit risk management in India. The results are encouraging and show high level of predictability, low bias while iterating the information. This study shows that information such as income level and default to ratio of CIBIL score are highly usable information by Banks to understand the default in commercial banks. The study also shows that the use of ANN yield good predictable result and can be used for credit risk management of a bank. The model is being able to learn properly and the results are consistent which mean that such a technique can be used in the long run for the credit risk management of a bank.