An artificial neural network-based expert system for loan Application evaluation at Kenya Commercial Bank

Abstract

A number of manual loan application evaluation models that use traditional judgmental methods have been used and continue to be used by commercial banks in Kenya. These systems however have shortcomings like more time taken to process loan applications and inconsistency in decision making by bank officials due to a variation of information provided by different customers. Some customers who are dissatisfied by this mode of processing loan applications subsequently moving to other banks or seek other financing modes. This poses a potential loss of business to a competitor commercial bank. The aim of this study was to analyze the current loan evaluation system at KCB with a view to design and develop an ANN architecture-based expert system for evaluating loans at the bank. The objectives of the study were: to find out the types of loans KCB offers to its clients; to examine the current systems used by KCB to evaluate loan applications; to determine the challenges faced when evaluating loan applications; to recommend suitable systems for improving the evaluation of loan applications; to design and develop an intelligent system that will improve loan applications evaluation process. This study was based on the expert system theory and the neural network architecture. Data were collected from information rich sources at KCB which involved sixteen respondents. The expert system was developed based on ANN architecture and modeled using the evolutionary development model. The neural network system was built using the back propagation algorithm. The developed expert system helps to fairly and uniformly evaluate loan applications efficiently and is considered an improvement to current loan evaluation processing. This system is recommended for use in Kenya Commercial Bank.