Student Clustering Based on Learning Behavior Data in the Intelligent Tutoring System

Literature Review

As it was previously mentioned, there are several research studies that deal with data-driven student clustering according to online learning behavior.

Mojarad et al. (2018) investigated data-driven student profiling in a web-based, adaptive assessment and learning system (ALEKS). The study grouped students into a set of clusters using data from the first half of the semester and 6 key characteristics: the initial assessment score percentage, the total number of assessments, the average days between assessments, the number of days since the initial assessment was taken, an average percentage score increase between assessments, and students’ final assessment score percentage in ALEKS (taken at the end of the class). By using Mean shift and K-means clustering algorithms, 5 distinct profiles were identified: strugglers, average students, sprinters, gritty, and coasters. The researchers found these profiles to be useful in enabling institutions and teachers to identify students in need and for subsequently devising and implementing appropriate interventions for groups of students with similar characteristics.