Asian Basic and Applied Research Journal

Car-loan is one of numerous services offered by creditors or financial institutions. Creditors are saddled with the responsibility of credit risk management on any loan facilities offered. To do this, they engage techniques and models to assess credibility of customers or car-loan applicants. The sole aim of this is to distinguish customers into two categories of credible customers and defaulters, which implies those that are more likely to pay their financial car-loan obligations and otherwise respectively. This will help to make inform decision in approving car-loans to customer and also help in retaining the sustainability of the creditor’s facility. Trailing to the aforementioned, this study made use of random forest algorithm, logistic regression and relevant dataset to develop two predictive models for car-loan applicant credibility. The developed predictive models for car-loan repayment credibility of customers were evaluated using accuracy, recall and precision. The results from this study revealed that both models were efficient in predicting the status of the car-loan credibility of customers while random forest performed relatively better than logistic regression.