Predicting Student Preferences for Digital and Traditional Loans: A Machine Learning Approach

Understanding student loan preferences is critical in designing inclusive and responsible financial services for the younger generation. However, many financial institutions and educational bodies still lack accurate predictive tools to identify students' borrowing tendencies. This study aims to predict Indonesian university students' preferences between digital and traditional loan services using machine learning algorithms, specifically Random Forest and Support Vector Machine (SVM). The dataset was collected through a structured online survey, capturing various demographic, academic, and behavioral financial features. Both models were evaluated using 5-fold cross-validation. The Random Forest model achieved an accuracy of 88.89%, while SVM reached 86.67%, demonstrating robust classification performance across both. Feature importance analysis revealed that loan purpose, personal beliefs regarding digital loans, spending behavior, saving habits, and parental income were key determinants of borrowing preference. These insights offer practical implications for loan providers, policymakers, and educational institutions in developing targeted interventions, financial literacy programs, and more ethical, data-informed lending strategies tailored to student needs.