Analysis of students' academic performance using LMS event logs

Modern educational process involves the use of electronic educational environments. These are special information systems that are both a means for storing educational materials and a tool for conducting tests, collecting homework, keeping a grade book, and working together. Such environments produce a large amount of data containing the recorded behavior of students and teachers within the educational process. This paper proposes an approach that allows one to analyze such data and discover typical student trajectories that lead to successful or unsuccessful learning outcomes. It is shown how process mining can be used to build models of the educational process based on the available data. We also show how you can evaluate the extent to which the synthesized model reflects the actual behavior of the system recorded in event logs. The paper contains not only a description of the proposed approach, but also a case study with its application to a real data set for an undergraduate educational program. It is clearly shown how, using our approach, it is possible to find out what factors lead to the formation of successful and unsuccessful student trajectories. The bottlenecks of the educational process were identified, as well as errors in the data, indicating the incorrect operation of the system. As a result of the analysis, points of special attention for administrators of the educational program were identified, as well as some signal events, the appearance of which in a student's individual trajectory can be an alarm. The application of the approach involves the use of free open source software, which further facilitates its deployment in a variety of educational organizations.