Е-Публикации
Технически университет - София

Abstract: Identifying at-risk students early is essential for improving educational outcomes, especially in resource-limited secondary schools that lack access to advanced AI systems due to financial constraints, limited technical infrastructure, insufficient human expertise, and fragmented or low-quality student data. Although accuracy values above 90% are often reported in the literature, such performances typically rely on intermediate-term grades (G1 and G2), large datasets, and highend computational resources. In realistic early-warning conditions where such intermediate grades are unavailable as in the present study accuracy levels between 60 - 72% are commonly observed. Our framework therefore aligns with the upper bounds of early-stage prediction accuracy while remaining fully compatible with low-resource environments. This paper presents a lightweight machine learning framework tailored for such environments to predict student performance and support early intervention. Five efficient machine learning algorithms, including Logistic Regression, Decision Tree, Random Forest, Gradient Boosting, and Naive Bayes, were evaluated using readily available student data. Logistic Regression achieved the best overall performance with 68.35% accuracy, 72.58% precision, 84.91% recall, and 0.7826 F1-score on the test set. The notably high recall ensures that most at-risk students are correctly identified. The system categorizes students into risk levels (High, Medium, Low), enabling strategic resource allocation. Key predictive factors include past failures, age, higher education aspiration, and social activity patterns. The framework operates offline on CPU-only hardware, requires less than 2 seconds of training time on standard laptops, and functions effectively with datasets of 300 - 400 students and basic teacher training (2-4 hours). This approach demonstrates that accessible AI tools can empower underserved schools to implement data-driven interventions with minimal infrastructure.

References

1. UNESCO, ªAi and education: Guidance for policy-makers,ºParis, France, 2021.
2. E. M. Allensworth and J. Q. Easton, ªWhat matters for staying on-track and graduating in chicago public high schools: A close look at course grades, failures, and attendance in the freshman year,ºConsortium on Chicago School Research, 2007.
3. M. A. Mac Iver, ªDropout early warning systems: A review of the literature,ºNational High School Center, 2012.
4. Kiketa, Vogel, H. Kashoba, S. Kasereka, P. Maniamfu, D. Kutangila, F. Buhendwa, S. Nyazabe, and J.-J. Katshitshi, ªDesign and implementation of a blended learning system for higher education in the democratic republic of congo as a response to covid-19 pandemic,ºInternational Journal of Emerging Technologies in Learning (iJET), vol. 17, no. 13, pp. 64±83, 2022.
5. C. Romero and S. Ventura, ªEducational data mining: a review of the state of the art,ºIEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 40, no. 6, pp. 601±618, 2010.
6. A. Y. Huang, O. H. Lu, and S. J. Yang, ªEducational data mining: A review from the perspective of big data,ºIEEE Access, vol. 8, pp. 129 967±129 983, 2020.
7. R. S. Baker and P. S. Inventado, ªBig data and education: Past, present, and future,ºJournal of Educational Technology & Society, vol. 22, no. 3, pp. 1±4, 2019.
8. B. Williamson, S. Bayne, and S. Shay, ªUnderstanding artificial intelligence in education: A critical review,ºLearning, Media and Technology, vol. 45, no. 3, pp. 252±265, 2020.
9. N. Selwyn, ªShould robots replace teachers? ai and the future of education,ºPolicy Press, 2020.
10. A. M. Shahiri, W. Husain, and N. A. Rashid, ªA review on predicting student’s performance using data mining techniques,ºProcedia Computer Science, vol. 72, pp. 414±422, 2015.
11. M. Hussain, W. Zhu, W. Zhang, and S. M. R. Abidi, ªStudent engagement predictions in an e-learning system and their impact on student course assessment scores,ºComputational Intelligence and Neuroscience, vol. 2018, 2018.
12. Z. Papamitsiou and A. A. Economides, ªLearning analytics: Principles and constraints,ºJournal of Educational Technology & Society, vol. 23, no. 1, pp. 1±14, 2020.
13. S. B. Kotsiantis, C. J. Pierrakeas, and P. E. Pintelas, ªPredicting students’ performance in distance learning using machine learning techniques,ºApplied Artificial Intelligence, vol. 18, no. 5, pp. 411±426, 2012.
14. A. I. Adekitan and O. Salau, ªThe impact of engineering students’ performance in the first three years on their graduation result using educational data mining,ºHeliyon, vol. 5, no. 2, p. e01250, 2019.
15. F. Okubo, T. Yamashita, A. Shimada, and H. Ogata, ªRecurrent neural networks for student performance prediction in massive open online courses,ºin Proceedings of the 2nd International Conference on Learning Analytics and Knowledge. ACM, 2017, pp. 506±510.
16. G. Gray, C. McGuinness, P. Owende, and A. Carthy, ªPredicting student performance using clickstream data and machine learning,ºInternational Journal of Information and Education Technology, vol. 4, no. 1, pp. 1±6, 2014.
17. J. Berens, K. Schneider, S. Gortz, È S. Oster, and J. Burghoff, ªEarly detection of students at risk±predicting student dropouts using administrative student data and machine learning methods,ºJournal of Educational Data Mining, vol. 11, no. 3, pp. 1±34, 2019.
18. L. Kemper, G. Vorhoff, and B. U. Wigger, ªPredicting student dropout: A machine learning approach,ºEuropean Journal of Higher Education, vol. 10, no. 1, pp. 28±47, 2020.
19. L. Wang, A. Sy, L. Liu, and C. Piech, ªDeep knowledge tracing on programming exercises,ºProceedings of the Fourth ACM Conference on Learning @ Scale, pp. 201±204, 2017.
20. A. Tlili, B. Shehata, M. A. Adarkwah, A. Bozkurt, D. T. Hickey, R. Huang, and B. Agyemang, ªA systematic review of research on artificial intelligence applications in higher education±where are the educators?ºInternational Journal of Educational Technology in Higher Education, vol. 20, no. 1, pp. 1±27, 2023.
21. S. Crossley, L. Paquette, M. Dascalu, D. S. McNamara, and R. S. Baker, ªCombining click-stream data with nlp tools to better understand mooc completion,ºProceedings of the sixth international conference on learning analytics & knowledge, pp. 6±14, 2016.
22. K. E. Arnold and M. D. Pistilli, ªCourse signals at purdue: Using learning analytics to increase student success,ºin Proceedings of the 2nd international conference on learning analytics and knowledge, 2012, pp. 267±270.
23. R. Aldmour, S. Yousef, T. Baker, and E. Benkhelifa, ªAn approach for offloading in mobile cloud computing to optimize power consumption and processing time,ºSustainable Computing: Informatics and Systems, vol. 31, p. 100562, 2021.
24. H. Crompton and D. Burke, ªArtificial intelligence in higher education: the state of the field,ºInternational Journal of Educational Technology in Higher Education, vol. 20, no. 1, p. 22, 2023.
25. M. Aldmour, R. Aldmour, A. Y. Al-Zoubi, and M. Sedky, ªOptimizing off-chain storage in blockchain of things systems: Implementing dockerized ipfs for enhanced efficiency,ºInternational Journal of Online & Biomedical Engineering, vol. 21, no. 1, 2025.
26. I. M. Castillo-MartÂınez, D. Flores-Bueno, S. M. Gomez-Puente, Â and V. O. Vite-Leon, Â ªAi in higher education: A systematic literature review,ºin Frontiers in Education, vol. 9. Frontiers Media SA, 2024, p. 1391485.
27. P. Cortez and A. M. G. Silva, ªUsing data mining to predict secondary school student performance,ºProceedings of 5th Annual Future Business Technology Conference, pp. 5±12, 2008.
28. F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg et al., ªScikit-learn: Machine learning in python,ºJournal of Machine Learning Research, vol. 12, pp. 2825±2830, 2011.
29. E. Commission, ªEthical guidelines on the use of artificial intelligence (ai) and data in teaching and learning for educators,ºBrussels, Belgium, 2022, accessed: 2025-10-20. [Online]. Available: https://education.ec.europa.eu/ethics-in-ai

Issue