Predicting Student On-Time Graduation Using Particle Swarm Optimization and Random Forest Algorithms

Higher education plays a crucial role in human resource development and national progress. A key indicator of educational quality is students' ability to graduate on time. Delays in graduation can lower the quality of higher education. Various academic and non-academic factors influence timely graduation rates. At Universitas Islam Syekh Yusuf, the trend of students graduating beyond the expected timeframe has risen over the past three years. However, the university lacks insight into the factors contributing to these delays. This research aims to identify factors causing delayed graduation using PSO and Random Forest to predict student graduation outcomes. The application of PSO reveals key factors influencing timely graduation, including study program, student active status, student leave of absence status, inactive status for semester 1, GPA1, and credit hours in semesters 1 and 2. Evaluation results show that using PSO and Random Forest to predict timely graduation achieves high accuracy (99.63%), precision (99.77%), recall (99.65%), and F1 score (99.71%).