Causal Analysis of Learning Performance Based on Bayesian Network and Mutual Information

Over the past few years, online learning has exploded in popularity due to the potentially unlimited enrollment, lack of geographical limitations, and free accessibility of many courses. However, learners are prone to have poor performance due to the unconstrained learning environment, lack of academic pressure, and low interactivity. Personalized intervention design with the learners’ background and learning behavior factors in mind may improve the learners’ performance. Causality strictly distinguishes cause from outcome factors and plays an irreplaceable role in designing guiding interventions. The goal of this paper is to construct a Bayesian network to make causal analysis and then provide personalized interventions for different learners to improve learning. This paper first constructs a Bayesian network based on background and learning behavior factors, combining expert knowledge and a structure learning algorithm. Then the important factors in the constructed network are selected using mutual information based on entropy. At last, we identify learners with poor performance using inference and propose personalized interventions, which may help with successful applications in education. Experimental results verify the effectiveness of the proposed method and demonstrate the impact of factors on learning performance.