An XGBoost-Based Knowledge Tracing Model

Abstract The knowledge tracing (KT) model is an effective means to realize the personalization of online education using artificial intelligence methods. It can accurately evaluate learning state students and conduct personalized instruction according characteristics different students. However, current models still have problems inaccurate prediction results poor features utilization. study applies XGBoost algorithm improve performance. In addition, also effectively handles multi-skill problem in by adding skills. Experimental show that best AUC value XGBoost-based reach 0.9855 multiple features. Furthermore, compared with previous used deep learning, saves more training time.