Investigating Students’ e-Book Reading Patterns with Markov Chains

Abstract

In this paper, we analyze students’ e-book reading patterns by using Markov Chains (MCs). We used click-stream data of 236 students while they read 7 different contents shared by the instructor across different weeks of the course. To analyze reading patterns, we first clustered students independently based on their interaction with each content. We grouped students in None, Low, Medium, and High clusters. Then by using MCs, we calculated cluster transition probabilities between different contents. We also visualized these patterns and applied a prediction algorithm to predict students’ reading patterns. Results revealed that students are likely to follow the same reading patterns across the semester. In other words, if a student reads less in the first content, s/he is likely to read less during the rest of the semester. We also found that transition data could be used to predict students’ further reading behaviors. The developed model can be used to plan an intervention system for at-risk students. Visualization of these transitions may help a teacher to understand how well students use contents.