Abstract

As people's demands grow, so does the need for loans to meet some of them. Every day, financial institutions get many loan applications. The financial industry faces significant challenges in managing loan defaults, which lead to increased fraud, bad debt, and financial losses. Specifically in Africa, financial institutions often lack accurate systems for assessing the creditworthiness of loan applicants.To address this issue, this study proposes a machine learning-based loan eligibility system that predicts the borrower’s creditworthiness .The methodology for this study was a combination of Design Science Research (DSR) and the Cross-Industry Standard Process for Data Mining (CRISP-DM) framework. The data was sourced from the Kaggle Dataset website, and the machine learning models were developed using Python and the Scikit-learn library. The results of the research indicate that the machine learning model developed was able to accurately predict creditworthiness. The Logistic Regression algorithm was used and evaluated using the performance accuracy metric which was at 80%. The system was  deployed as a web-based application using the Streamlit framework. Future research and improvement such as data enrichment, advanced analytics, and financial regulatory compliance are however still required.