METHODOLOGY FOR ANALYSING LMS DATA TO PREDICT STUDENT DROPOUT RISK IN HIGHER EDUCATION

Nowadays, educational institutions use Learning Management Systems (LMS) to support students in the learning process. LMS technical data analysis enables the monitoring of student activities and early identification of those at risk of failing a course. Data gathered during the educational process facilitates the adaptation of learning content to meet each student's individual needs. By leveraging this data, institutions can implement adaptive education, allowing study programs to be structured based on personalized learning pathways, intelligent recommendation systems, and dynamic curriculum adjustments. Additionally, by analyzing student model data, it is possible to assess dropout risks. As a result, research on student attrition rates has gained increased attention. This paper examines the methodology for analyzing Moodle LMS data to adaptively detect factors influencing student dropout risk. The research explores the potential of analyzing log file data generated by Moodle LMS to identify student model parameters and their impact on student success throughout the entire educational process. By utilizing learning patterns and engagement indicators, activity log data from more than seven hundred students at Riga Technical University's Moodle e-learning system was analyzed. The research aimed to identify correlations and relationships between several factors, including the availability of resources for students, the number of graded activities, activity types, views, and other relevant data. By analyzing correlations between fluctuations in students' learning achievements and behavioral patterns in e-learning platforms, the study aims to identify key indicators and metrics for predicting dropout tendencies. The findings suggest that a decline in engagement, the presence of negative patterns, or the absence of consistent learning behaviors serve as reliable indicators of students at risk of dropping out.