New Perspectives in Science Education

This study explores the relationship between student behavioural traits and academic achievement using machine learning techniques. While the learning styles outlined in the VARK model (visual, auditory, reading/writing, kinesthetic) offer insights into student preferences, they only cover one aspect of behavioural traits. In this context, behavioural characteristics refer to a broader range of observable actions and tendencies displayed by students during the learning process. These include interaction patterns with content, collaboration styles, data entry methods, and engagement with the learning management system (LMS) on the Moodle platform. Using experimental data gathered from students taking part in a programme supported by the Estonian National Commission for Professional Development, this study applies decision tree and random forest algorithms to analyse both declared learning style preferences and behavioural data derived from Moodle logs. This research aims to assess the predictive capacity of these traits and their influence on academic success. The findings indicate that, while learning styles alone have limited predictive value, a detailed analysis of behavioural characteristics, work preferences, and adaptability offers a stronger foundation for forecasting academic performance in short-term learning environments. These outcomes underscore the potential of data-driven educational approaches to enhance curriculum design, tailor learning experiences, and promote lifelong learning.