Student Performance Prediction via Attention-Based Multi-Layer Long-Short Term Memory

Online education has attracted a large number of students in recent years, because it breaks through the limitations time and space makes high-quality at your fingertips. The method predicting student performance is to analyze predict student’s final by collecting demographic data such as gender, age, highest level, clickstream generated when interact with VLE different types specific courses, which are widely used online platforms. This article proposes model via Attention-based Multi-layer LSTM (AML), combines for comprehensive analysis. We hope that we can obtain higher prediction accuracy soon possible provide timely intervention. results show proposed improve 0.52% - 0.85% F1 score 0.89% 2.30% on four-class classification task well 0.15% 0.97% 0.21% 2.77% binary from week 5 25.