A Learning Early-Warning Model Based on Knowledge Points
Jiahe Zhai, Zhengzhou Zhu, Deqi Li, Nanxiong Huang, Kaiyue Zhang, Yuqi Huang
Learning early-warning is one of the important ways to realize adaptive learning. Aiming at the problem of too large prediction granularity in learning early-warning, we divide student’s characters into three dimensions (knowledge, behavior and emotion). Secondly, we predict the student’s master degree of knowledge, based on the knowledge point. And then we realized learning early-warning model. In the model, we take 60 points as the learning early-warning standard, and take RF and GDBT as base classifiers, and give the strategy of selecting the basic model. The experiment shows that the prediction of knowledge mastery of the model and the real data Pearson correlation coefficient can reach 0.904279, and the prediction accuracy of the model below the early-warning line can reach 76%.
The final publication is available at Springer via https://doi.org/10.1007/978-3-030-22244-4_1