Kuwait-University-Journal-of-Law-header
Search
Journal of Law

Previous Issues

Advance Search
Year : From To Vol
Issue Discipline:
Author

Volume :3 Issue : 2 1996      Add To Cart                                                                    Download

Predicting the Credit Viability of Bank Customers: Discriminant Analysis Versus Neural Networks

Auther : Masood A. Badri
Ahmad Al-Mutawa
Aquel M. Hadi

This paper compares the use of artificial neural networks (ANN) and multiple discriminant analysis (MDA) to predict bank customer’s credit risk. Traditionally, MDA is regarded as the most applicable statistical method for the prediction of credit risk and is widely used by financial institutions and other types of organizations. However, MDA is based on a linear equation and therefore subject to the limitation of not being appropriate for application to problems that include the analysis of non-linear data. In addition, most financial applications of MDA ignore the fact that certain assumptions with regard to the independent variables are required; thus, resulting in methodologically wrong outcomes. ANN, on the other hand, most frequently incorporate both linear and non-linear functional components and are, therefore, more suitable for problems that include the analysis of a greater variety of data types. A financial application (bank customer’s credit scoring) is presented as a case, to measure the comparative performance of the two methods. Results indicate that ANN can outperform MDA in forecasting credit risk.

Journal of Law
Journal of Law

You are Visitor No.

77787

Journal of Law
Tell your friendsJournal of Law
Journal of Law
Journal of Law

Last Updated

Nov 19, 2019

Journal of Law
Journal of Law
Journal of Law

Please enter your email Here to receive our news

Journal of Law