Predicting student grades via adaptive multi-level learning models
Ali Reza Ibrahimzada, Kerem Kosif, Ahmed Said Gulsen, Yavuz Selim Yaglica, Ali Cakmak · 2026
Read Predicting student grades via adaptive multi-level learning models by Ali Reza Ibrahimzada, Kerem Kosif, Ahmed Said Gulsen, Yavuz Selim Yaglica, Ali Cakmak (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13478600.txt.
Description
Ibrahimzada and co-authors present a grade prediction framework in 2026 that splits student-course records into similar groups and lets a specialist model serve each group. Predicting exact letter grades before a course starts can guide advising, but one global model fits varied courses poorly. The framework accepts many model types and includes a fallback that merges sparse slices into local validation pools when data run thin.
On one dense, traditional university dataset, the framework cut prediction error by 18.6 percent against global baselines, with tree ensembles winning most groups. On a sparse, volatile dataset it chose neural models for 48 percent of groups and collaborative filtering, which borrows patterns from similar students, for 22 percent of time windows. Significance testing returned an extremely small p-value, and the authors claim the system shifts tuning burden from hand rules to self-correcting selection across institutions.
Catalog record
| CREATOR | Ali Reza Ibrahimzada, Kerem Kosif, Ahmed Said Gulsen, Yavuz Selim Yaglica, Ali Cakmak |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 106 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |