Presentation Schedule
Predicting a New Student’s Math Score Based on the Scores of Previous Students (100359)
Session Chair: Fumiyo Seimiya
This presentation will be live-streamed via Zoom (Online Access)
Wednesday, 7 January 2026 16:35
Session: Session 2 (Parallel)
Room: Live-Stream Room 3
Presentation Type: Live-Stream Presentation
– click here to convert to your timezone
Accurately predicting students’ academic performance is a growing area of interest in educational data mining and learning analytics, with the potential to inform policy decisions, resource allocation, and personalized interventions. This study explores the feasibility of predicting a new student’s mathematics score based on the performance patterns of previous student cohorts. The central idea is that past achievement within the same institutional and instructional context may provide valuable signals for anticipating outcomes of incoming students.
Using historical mathematics scores from multiple cohorts across several academic years, we investigate different modeling strategies, including baseline comparisons and machine learning approaches such as linear regression and random forest. Predictors incorporate not only cohort-level trends but also contextual features such as teacher tenure, exam year, and school characteristics. Evaluation metrics (RMSE, MAE, R²) and time-based validation are applied to ensure robustness and prevent data leakage.
Preliminary results indicate that incorporating cohort-level features improves prediction accuracy beyond simple baselines, though variation across subgroups highlights the importance of fairness considerations. We also discuss the ethical implications of predictive analytics in education, emphasizing that predictions should support early identification and resource allocation rather than reinforce stereotypes or high-stakes tracking.
By framing the problem at the intersection of cohort-level modeling and individual-level forecasting, this work contributes to ongoing discussions in predictive educational analytics. The study underscores both the potential and limitations of leveraging prior cohorts’ data to inform expectations for new students, pointing toward more nuanced, equity-aware applications of predictive modeling in educational practice.
Authors:
Iyad Suleiman, Tel Hai Academic College, Israel
Rozan Abbas, Emek Yizrael Academic College, Israel
Amani Attallah, Yizrael Academc Collge, Israel
Rula Jiryis, Kinneret Academic College, Israel
About the Presenter(s)
Dr Iyad Suleiman is a Lecturer in Tel Hai Academic College
Connect on Linkedin
https://www.linkedin.com/in/iyad-suleiman-62100943/
See this presentation on the full schedule – Wednesday Schedule
Conference Comments & Feedback
Place a comment using your LinkedIn profileShare this Presentation
Comments
Powered by WP LinkPress