Event date · · TRACE

Jointly Predicting Courses and Grades Using a Transformer-Based Model

FACT STATEMENT

A paper introduces TRACE, a transformer-based model that jointly predicts courses and grades for an upcoming semester. Trained on ten years of institutional data, it reduces mean absolute error by nearly 50% compared to an identical architecture predicting grades alone, and outperforms LSTM-based and graph neural network-based approaches.

What happened

Existing predictive models in learning analytics often treat student academic history as a simple sequence, overlooking the concurrent nature of courses taken within a semester. This paper introduces TRACE, a TRansformer for Academic Course-grade Estimation, which jointly predicts both the set of courses a student will take and their corresponding grades for an upcoming semester. The approach encodes courses on a per-semester basis to capture course concurrency and uses a novel loss function combining course-set prediction with grade prediction. Trained on ten years of institutional data, the joint prediction model reduces mean absolute error by nearly 50% compared to an identical architecture that predicts grades alone, and outperforms traditional LSTM-based sequential models and graph neural network-based approaches.

Technical significance

The model encodes courses per semester to capture concurrency effects, and uses a combined loss for course-set and grade prediction. This joint objective leads to a nearly 50% reduction in mean absolute error versus grade-only prediction, and outperforms LSTM and GNN baselines.

Industry impact

The approach could improve academic advising and early intervention systems by providing more accurate predictions of student course loads and performance, especially for students with heavy or challenging schedules.

Decision value

Institutions could use TRACE to enhance student success initiatives, optimize course planning, and allocate support resources more effectively, potentially improving retention and graduation rates.

What to watch

Next signals include validation on external institutional datasets, integration into learning management systems, and extension to other prediction tasks such as dropout risk or major selection.

DECISION BRIEF

Turn the evidence into a decision.

See how AIGC.NEWS separates verified change, judgment, and the next signal to watch.