Data-Driven Analysis of Student Engagement in Time-Limited Computer Laboratories

Computer laboratories are learning environments where students learn programming languages by practicing under teaching assistants' supervision. This paper presents the outcomes of a real case study carried out in our university in the context of a database course, where learning SQL is one of the main topics. The aim of the study is to analyze the level of engagement of the laboratory participants by tracing and correlating the accesses of the students to each laboratory exercise, the successful/failed attempts to solve the exercises, the students' requests for help, and the interventions of teaching assistants. The acquired data are analyzed by means of a sequence pattern mining approach, which automatically discovers recurrent temporal patterns. The mined patterns are mapped to behavioral, cognitive engagement, and affective key indicators, thus allowing students to be profiled according to their level of engagement in all the identified dimensions. To efficiently extract the desired indicators, the mining algorithm enforces ad hoc constraints on the pattern categories of interest. The student profiles and the correlations among different engagement dimensions extracted from the experimental data have been shown to be helpful for the planning of future learning experiences.