Document Details

Document Type : Thesis 
Document Title :
Model Development Using Neural Networks for Solution of Car Purchase Credit Problem in Saudi Arabia
تطوير نموذج باستخدام الشبكات العصبية لحل مشكلة الثقة لعملية شراء السيارات في المملكة العربية السعودية
 
Subject : Faculty of Engineering 
Document Language : Arabic 
Abstract : Many companies turned to credit purchase business. Among those are car agents. Although there is a risk in such deals, but it is better than piling up goods which encounter losses. Thus, the credit scoring systems appeared to help in limiting the risks originating from credits. Credit scoring is a method by which every applicant is scored depending on a set of features and attributes he posses. This score is derived from the behavior of old applicants having similar attributes. Recently, Neural Networks were involved in this issue to give precise weight for each value of different factors (features) used in the modeling process. Many software programs are available in the market for modeling the credit scoring problem such as NeuroSolution, which is used in this research. This research applies NN models for the credit scoring problem in car purchasing using data obtained from a local car agent. The most difficult stage in this research was the data collection phase. Most companies consider their data as confidential. At the same time, the data available is not sufficient because many important factors are not included in the application forms. The research used four NN models suggested by previous researchers in the same field. The most appropriate NN model was the Radial Basis Function (RBF). The results show slight improvement to the current system obtained by the company. The modest performance of the model may be referred to the insufficiency of data size and the unavailability of some important factors which are usually considered in the credit scoring problem. This necessitates further research on the issue and points out the need for paying more attention to the documentation of features and behavior of past customers, in order for the company to limit the risk of granting credit for unworthy applicants. 
Supervisor : Dr. Abdullah Bafail 
Thesis Type : Master Thesis 
Publishing Year : 1430 AH
2009 AD 
Co-Supervisor : Dr. Adnan Fakieh 
Added Date : Sunday, December 27, 2009 

Researchers

Researcher Name (Arabic)Researcher Name (English)Researcher TypeDr GradeEmail
أسامة عبد الكريم منصورMansour, Osama Abdul-KareemResearcherMaster 

Files

File NameTypeDescription
 24524.pdf pdf 

Back To Researches Page