Abstract:
The very nature of the banking business is so sensitive because more than 85% of their liability is
deposits from customers. Banks use these deposits to generate credit for their borrowers, which in
fact is a revenue generating activity for most banks. However, this credit creation process exposes
Banks to high default risk, which might lead to financial distress including bankruptcy. So banks
need to manage credit risk inherent in the entire portfolio as well as the risk in individual credits or
transactions. In this study therefore, a case-based credit approval decision making knowledge
base system that uses data mining results is proposed by applying empirical research design. The
researcher used manual and automated knowledge acquisition techniques, such as interview,
document analysis and data mining. To identify the best prediction model for Credit approval
decision making, three experiments using three classification algorithms were conducted. Finally,
the researcher decided to use the results of J48 decision tree classification algorithm in the
development of the prototype case-base System because it registered better performance than other
classifiers. The developed model was tested with test instances and only those instances registers
more than 95% accuracy were used to develop a knowledge base for the CBR development for a
better efficiency. Then, the implementation of the prototype using JCOLIBERI version 1.1 which
is object oriented case-based reasoning framework is realized. Finally, testing of the prototype
case-based reasoning system is done to evaluate the performance of the system. The prototype is
evaluated using system testing and user acceptance testing. Testing system performance in terms
of precision, recall and f-measure registered 83%, 73 % and 77 %, respectively. Also user
acceptance testing achieved 83.2% performance. The evaluation of the prototype shows a
promising result to design an applicable intelligent system that supports effective and efficient
credit approval decisions making. But, the current system suggests no explanation about the
correct action to be taken; as a result a hybrid explanation driven system by combining Case based
reasoning with Rule based reasoning is recommended as a future research direction.
