MOOC Performance Prediction and Online Design Instructional Suggestions Based on LightGBM

Yimin Ren, Jun Wang, Jia Hao, Jianhou Gan, Ken Chen

Abstract: Recently, the research on the teaching effectiveness of online teaching has gradually become the focus of people’s attention. As a well-known learning platform, MOOC has also become the main front for many learners to conduct online learning. However, some students are not clear about their own learning situation during the learning process, so that they can’t get a qualifying grade in a MOOC course. Thus, In order to make the teachers and learners to anticipate the learning performance and then check the gaps as early as possible, we propose a MOOC performance prediction model based on the algorithm of LightGBM, and then compare its results with the others state-of art machine learning algorithms. Experimental results show that our proposed method outperform than the others. Research also leverages LightGBM’s interpretability, Analyzed the external environment and existing cognitive structure, two factors that affect online learning. And this is the basis for suggestions on online instructional design.