Investigating Subpopulation of Students in Digital Textbook Reading Logs by Clustering

Abstract

The increasing volume of student reading logs from virtual learning environment (VLE) provides opportunities for mining student’ engagement pattern in digital textbook reading. In order to mine and measure students’ engagement pattern, in this paper, we extract several students’ reading interaction variables from the digital textbook as metrics for the measurement of reading engagement. Moreover, in order to explore the presence of subpopulation of students that can be differentiated based on their engagement patterns and academic performances, we cluster students into different groups. Students are clustered based on their reading interactions such as total session of reading, total notes adding, etc. Accordingly, we identify students’ engagement patterns from different groups based on the clustering analysis results. Several student subpopulations such as low engagement high academic performances and low engagement low academic performances are identified based on students’ reading interaction characteristics by clustering analysis. The obtained results can be used to provide researchers with opportunities to intervene in the specific group of students and also an optimal choice for student grouping.