Researchers have developed a new Transformer-based model called TRACE (TRansformer for Academic Course-grade Estimation) to improve predictions of student academic performance. Unlike previous models that treated academic history as a simple sequence, TRACE accounts for the concurrent nature of courses taken within a semester. By jointly predicting both the courses a student will take and their corresponding grades, the model significantly reduces prediction error compared to models that only predict grades or use traditional LSTM and GNN approaches. This new method, trained on ten years of institutional data, shows potential for integration into early detection systems in higher education. AI
IMPACT This model could enhance early detection systems in educational institutions by providing more accurate student performance predictions.
RANK_REASON This is a research paper detailing a new model and its performance on academic prediction tasks. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- CatalyzeX
- DagsHub
- Gotit.pub
- graph neural network
- Hugging Face
- long short-term memory
- ScienceCast
- TRACE
- Transformer++
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
