Old-School Trees Beat Attention Models in Student Performance Prediction, Rigorous Benchmark Finds
A leakage-aware benchmark of seven models across three educational datasets found that Gradient Boosting outperformed attention-based neural networks, which showed ...
