2017/08/14 by Christian Hansen, Hansen, Christian, Casper Worm Hansen +7
Computer Science · #Computers and Society (cs.CY) #FOS: Computer and information sciences #Human-Computer Interaction (cs.HC) #Intelligent Tutoring Systems and Adaptive Learning #Online Learning and Analytics #Software System Performance and Reliability
paper · pdf · doi:10.48550/arxiv.1708.04164
openalex publication_date 2017/08/14 · openalex created_date 2022/09/15 · openalex updated_date 2026/07/28
The analysis of log data generated by online educational systems is an\nimportant task for improving the systems, and furthering our knowledge of how\nstudents learn. This paper uses previously unseen log data from Edulab, the\nlargest provider of digital learning for mathematics in Denmark, to analyse the\nsessions of its users, where 1.08 million student sessions are extracted from a\nsubset of their data. We propose to model students as a distribution of\ndifferent underlying student behaviours, where the sequence of actions from\neach session belongs to an underlying student behaviour. We model student\nbehaviour as Markov chains, such that a student is modelled as a distribution\nof Markov chains, which are estimated using a modified k-means clustering\nalgorithm. The resulting Markov chains are readily interpretable, and in a\nqualitative analysis around 125,000 student sessions are identified as\nexhibiting unproductive student behaviour. Based on our results this student\nrepresentation is promising, especially for educational systems offering many\ndifferent learning usages, and offers an alternative to common approaches like\nmodelling student behaviour as a single Markov chain often done in the\nliterature.\n