This paper presents the design, development, and evaluation of an AI-powered student performance prediction system aimed at proactively identifying at-risk students within an academic semester. The system integrates a microservices-based web architecture utilizing React.js, Node.js, and MongoDB with Python-based machine learning microservices implementing Random Forest algorithms for continuous score regression and binary pass/fail classification. A novel weighted scoring formula incorporates academic and extracurricular metrics, enabling a holistic evaluation of student engagement. The system provides automated risk categorization, real-time dashboards for faculty, and rapid inference performance. Experimental results demonstrate high predictive accuracy (94.2% classifier accuracy, R²=0.88 regression), confirming the efficacy of the ensemble learning approach and the microservice architecture in educational settings. This work bridges the gap between offline predictive models and actionable, real-time educational analytics, supporting a shift from reactive grading to proactive intervention.
The AI-Powered Student Performance Prediction System represents a significant advancement in Educational Technology by synthesizing predictive machine learning models within decoupled full-stack architectures. The integration of targeted machine learning models within decoupled full-stack architecture demonstrates that complex machine learning inference can be seamlessly integrated into highly responsive educational web applications. The mathematical formulation of comprehensive academic engagement successfully captures holistic performance, balancing traditional exams with practical projects and extracurricular initiatives.
The implementation of Random Forest and Gradient Boosting algorithms has yielded highly accurate predictions for both continuous scores and binary outcomes, validating ensemble learning as an optimal approach for structured academic data. Most importantly, automated risk assessment logic transforms raw statistical predictions into actionable intelligence, empowering academic institutions to shift their pedagogical strategies from retroactive grading to proactive intervention, ensuring struggling students receive targeted support before reaching the point of failure. As educational institutions continue to generate unprecedented volumes of data, the continued development of ethical, interpretable, and equitable predictive systems remains paramount. Future research should prioritize fairness-aware modeling, cross-institutional validation, deep learning for temporal pattern analysis, and seamless integration with existing educational infrastructure while maintaining commitment to student privacy, algorithmic transparency, and pedagogical integrity.
References
[1]L. Corrin, “Supporting the use of student-facing learning analytics in the classroom,” in Routledge, pp. 208–220, 2018.
[2]P. Kaushik, M. Kakkar, M. Yadav, C. Jegadheesan, P. N. Sripada, and K. Chahal, “Data analytics in education: Enhancing student learning outcomes,” in Proc. IEEE ICAC2N, pp. 1542–1547, 2024.
[3]V. Kovanovic, C. Mazziotti, and J. Lodge, “Learning analytics for primary and secondary schools,” J. Learn. Anal., vol. 8, no. 2, pp. 1–5, 2021.
[4]O. Ajuwon, E. Animashaun, and N. Chiekezie, “Integrating AI and technology in educational administration: Improving efficiency and educational quality,” Open Access Res. J. Sci. Technol., vol. 11, no. 2, pp. 116–127, 2024.
[5]E. Agyemang, J. Mensah, O.-A. Ampomah, L. Agyekum, J. Akuoko-Frimpong, A. Quansah, and O. Akinlosotu, “Predicting students’ academic performance via machine learning algorithms: An empirical review and practical application,” Comput. Eng. Intell. Syst., vol. 15, no. 1, 2024.
[6]A. Garg, N. B. Garg, P. Ghosh, A. Bansal, U. K. Lilhore, and S. Simaiya, “A machine learning-based automatic model to predicting performance of students,” in Proc. IEEE CCET, vol. 17, pp. 1–5, 2022.
[7]P. J. B. Pajila, G. Anand, K. Neela, M. Gowri, M. Karthikeyan, and R. Nithyanandhan, “Academic performance evaluation using data engineering and multi-model machine learning system,” in Proc. IEEE ICSTSDG, pp. 1–6, 2024.
[8]S. Elatia and D. Ipperciel, “Learning analytics and education data mining in higher education,” in IGI Global, pp. 108–126, 2021.
[9]S. K. G. and M. Kurni, “Educational data mining and learning analytics,” in Springer, pp. 29–60, 2021.
[10]S. Alarcón-Loza, D. Calderón-Onofre, K. Mite-Baidal, and M. Macías-Plúas, “Design of a predictive model to evaluate academic risk using data mining,” in Springer Nature Switzerland, pp. 221–235, 2023.
[11]S. M. Jayaprakash, E. W. Moody, E. J. M. Lauría, J. R. Regan, and J. D. Baron, “Early alert of academically at-risk students: An open source analytics initiative,” J. Learn. Anal., vol. 1, no. 1, pp. 6–47, 2014.
[12]K. D. Gori, K. Zalawadia, and H. Bhaidasna, “Explainable AI for students performance prediction system,” in Proc. IEEE PuneCon, pp. 1–6, 2025.
[13]A. Efendi, I. Fitri, and G. W. Nurcahyo, “Improvement of machine learning algorithms with hyperparameter tuning on various datasets,” in Proc. IEEE ICFTSS, vol. 6, pp. 75–79, 2024.
[14]N. S. Ahmed and M. Hikmat Sadiq, “Clarify of the random forest algorithm in an educational field,” in Proc. IEEE ICOASE, pp. 179–184, 2018.
[15]N. Nazeer, “Operationalizing AI: The central role of MLOps in deploying scalable ML systems,” J. Inf. Syst. Eng. Manage., vol. 10, no. 62s, pp. 599–607, 2025.
O. Chaplia and H. Klym, “Designing a Node.js project architecture for RESTful microservices,” in Proc. IEEE IDAACS, vol. 17, pp. 808–811, 2023.
How to Cite This Paper
Naveen Kumar S, Abishek K, Karthikeyan V, Iyyappan K, Sudhakar P (2026). AI-Powered Student Performance Prediction System using Machine Learning, and Predictive Analytics in Higher Education. International Journal of Computer Techniques, 13(2). ISSN: 2394-2231.
