A Machine Learning Approach to Predicting On-Time Graduation in Indonesian Higher Education

Graduating on schedule is a critical milestone for students in higher education, serving as a key indicator of both institutional effectiveness and student success. This study uses machine learning techniques to predict on-time graduation in Indonesian higher education. A dataset comprising 133 students from an engineering department over four academic years (2019–2023) was analyzed using the CRISP-DM framework. The research employed nine machine learning models, including Random Forest, Logistic Regression, Neural Networks, etc., to identify key predictors of on-time graduation. The result showed that Random Forest outperformed other models by achieving an accuracy of 85% and an AUC of 0.875. Additionally, the study developed a learning analytics dashboard to visualize predictive insights, offering actionable data for educators and administrators. The system’s performance was evaluated based on functionality, usability, efficiency, and reliability as the key intersecting factors from ISO/IEC 25010 and WebQEM frameworks, validating its quality and relevance for practical educational use. The result demonstrated high functionality, efficiency, and reliability, and positive usability feedback was received from both students and educators. The findings highlight the top ten important factors, such as cumulative GPA (CGPA), extracurricular involvement, programming, and social science courses, that predict on-time graduation, providing valuable insights for enhancing student outcomes in Indonesian higher education.

Recommended citation: Pawitra, M. A. S., Hung, H. C., & Jati, H. (2024). A Machine Learning Approach to Predicting On-Time Graduation in Indonesian Higher Education. Elinvo (Electronics, Informatics, and Vocational Education), 9(2), 294-308. https://doi.org/10.21831/elinvo.v9i2.77052
Download Paper