ASSESSING THE CREDITWORTHINESS STATUS OF MOBILE PHONE USERS USING SUPPORT VECTOR MACHINE

Abstract

Credit risk is a major concern to lenders, it is important for any lending company to be able to determine when to approve and when to decline a loan. Machine learning techniques have recently been adopted to help identify defaulting customers, and also help to speed up the decision-making process of approving a loan. In this study, relevant features that are related to customers' credit scoring are selected, and we made use of a support vector machine to build a model that could solve the underlying problems. From the test result, our developed model could predict a borrower's compliance status to loan payment. The model was able to attain a performance measure for Precision, recall, accuracy, and f1-score on test data with values of 97.2 %, 100.0 %, 99.1 %, and 98.6 % respectively. This indicates that the Support vector machine is an effective approach that could be used in credit scoring, and our developed model
can be classified as a good classification.