| ISSN: 2583-4371 | ISSN: 2583-4371JOURNAL ARTICLE
Abstract: Lecturers in large classes often identify struggling students late because monitoring relies on manual observation. This paper presents SPARK (Student Performance Analysis and Risk Knowledge), a machine learning early warning system that classifies students as High, Medium or Low academic risk from the grades available during a semester and delivers the results through an interactive Streamlit dashboard. Risk is defined from the final grade and estimated at two points: after the first period grade (Stage 1) and after the second (Stage 2). Six classifiers (Decision Tree, Random Forest, KNN, Logistic Regression, SVM and XGBoost) and a rule baseline were evaluated on 316 development students of the UCI Student Performance (Mathematics) data, a sealed hold-out set of 79 students, and 275 non-overlapping Portuguese-language students for cross-subject testing. The High-risk threshold was set to reach a recall of 0.90. The Random Forest, chosen by a rule fixed before testing, caught 24 of 26 High-risk hold-out students without labelling any as Low risk; on Portuguese students its recall fell to 0.83 and false alarms rose. A rule that applies the risk bands to the latest grade performed almost as well as every model, showing that the first two grades carry most of the usable signal. Thresholds did not transfer across subjects and should be calibrated per course.
Keywords: early warning system, student risk prediction, learning analytics, machine learning, class imbalance, dashboard.
Article Info: Received: 05 Oct 2026, Received in revised form: 02 Nov 2026, Accepted: 07 Nov 2026, Available online: 12 Nov 2026
[1] P. Cortez and A. Silva, “Using data mining to predict secondary school student performance,” in Proc. 5th Future Bus. Technol. Conf. (FUBUTEC), A. Brito and J. Teixeira, Eds. EUROSIS, 2008, pp. 5–12. [Online]. Available: https://archive.ics.uci.edu/dataset/320/student+performance
[2] M. N. Gul, W. Abbasi, M. Z. Babar et al., “Data driven decisions in education using a comprehensive machine learning framework for student performance prediction,” Discover Comput., vol. 28, Art. no. 153, 2025, doi: 10.1007/s10791-025-09585-3.
[3] S. Wiyono and T. Abidin, “Comparative study of machine learning KNN, SVM, and Decision Tree algorithm to predict student’s performance,” Int. J. Res. GRANTHAALAYAH, vol. 7, no. 1, pp. 190–196, 2019, doi: 10.29121/granthaalayah.v7.i1.2019.1048.
[4] J. Wang and Y. Yu, “Machine learning approach to student performance prediction of online learning,” PLOS ONE, vol. 20, no. 1, Art. no. e0299018, 2025, doi: 10.1371/journal.pone.0299018.
[5] Hamsiah, N. Adiyati, and R. Subekti, “Early prediction of at risk students using minimal data: A machine learning framework for higher education,” Digitus: J. Comput. Sci. Appl., vol. 3, no. 2, 2025, doi: 10.61978/digitus.v3i2.953.
[6] M. El Jihaoui, O. E. K. Abra, and K. Mansouri, “Towards a fair and transparent early warning system for identifying students at risk of academic failure,” Front. Educ., vol. 11, 2026, doi: 10.3389/feduc.2026.1878686.
[7] B. Pei, Y. Cheng, A. Ambrose, E. Dziadula, W. Xing, and J. Lu, “LearningViz: A dashboard for visualizing, analyzing and closing learning performance gaps, a case study approach,” Smart Learn. Environ., vol. 11, Art. no. 56, 2024, doi: 10.1186/s40561-024-00346-1.
[8] H. B. Santoso, A. K. Batuparan, R. Y. K. Isal, and W. H. Goodridge, “The development of a learning dashboard for lecturers: A case study on a student-centered e-learning environment,” J. Educators Online, vol. 15, no. 1, 2018. [Online]. Available: https://eric.ed.gov/?id=EJ1168957
[9] Danyaro, K. U., Nasser, M., Zakari, A., Abdullahi, S., Khanzada, A., Yakubu, M. M., & Shoaib, S. (2025). LLM-based code generation: a systematic literature review with technical and demographic insights. IEEE Access, 13, 194915-194939.
[10] Nasser, M. (2024). Personalized learning through AI: enhancing student engagement and teacher effectiveness. International Journal of Teaching, Learning and Education, 3(6), 23-26.
[11] Nasser, M., Yusof, U. K., & Salim, N. (2023). Deep learning based methods for molecular similarity searching: a systematic review. Processes, 11(5), 1340.
[12] L. Graf, P. Bassner, M. Anzinger, F. Dietrich, S. Krusche, and O. Poquet, “Interactive learning dashboards: Rethinking learning visualisations as engagement tools,” Educ. Inf. Technol., 2026, doi: 10.1007/s10639-026-14082-1.
