Loan Risk Prediction Model Using Machine Learning رسالة ماجستير

Abstract

With the increase in the number of Credit applications and current expansion in loan and financing services, processing these applications manually has become time-consuming. Customers may apply for loans, such as establishing startups, marriage, buying new or old cars, mortgage, education, new construction projects, etc. However, only qualified customers can be granted loans to ensure the funds are going to be returned. Many banks around the world still use manual lending decision-making procedures to make judgements. These decisions are made by bank officers based on personal interaction with loan applicants and customer transactional lending profiles. Bank loan officers make their credit decisions based on information about applicants if they are qualified, such as personal impressions about loan applicants and other information not existent in financial records. Making decisions requires a lot of efforts, time, as well as deep knowledge about customers and their loan plans. Bank loan officers also use transactional lending practice to choose their quantitative information, such as cash flows, annual reports, and liquidity measures. They combine personal information and quantitative information to set the bank officers' final decision or recommendation. This study aims to utilize Machine Learning (ML) to create an automated decision support system to identify customers who are eligible to receive loans based on quantitative information. III Supervised Machine Learning methods will be used to judge whether the customer will pay back. These methods will use training datasets from existing customer records to build a classification model; the training dataset is manually judged. The subject of this study is the Bank of Palestine, which provided us with the necessary training data to carry out this research. The outcome of the research demonstrated promising decisions based on quantitative information; these judgement results accompanied with Bank officers’ personal interactions with the client will definitely reduce the time spent in loan decision making with high accuracy similar to the manual workflow implemented by the Bank of Palestine