Machine Learning-Based Identification and Behavioral Analysis of Student Learning Styles in Online Learning Environments Using FSLSM and Clustering Techniques

Abstract

A major objective of the systematic education process is lifelong learning. Monitoring and identification of the learning styles of the learners in delivering educational content plays a significant role for improving both teaching and learning aspects. In E-learning environment, identification of the learning style plays a pivotal role as it highly impacts the module content design process for the online learning environment to improve the knowledge of the students in a precise manner. When it is considered the Moodle learning platform, a popular method used for learning style identification is considering the access frequency. But considering only the access frequency will always not predict the correct results as there may be situations where students access the course material online but do not spend sufficient time with educational materials. The proposed methodology is a novel improved approach which considers both access frequency and the time spent on the learning material by applying the K-Means Clustering. In this research, Felder Silverman learning Style model (FSLSM) which is an acceptable learning style identification methodology was used as the leaning model. For the data collection process, it has been used the Moodle logs for the getting the access frequency and a deigned reusable Moodle plugin for measuring the total time spent on the activities. Then K- Means clustering was applied, which is an unsupervised machine learning algorithm to identify the learning style of the learners. The development of an application that employs these models to dynamically identify and predict student learning styles adds practical value, enabling integration into learning management systems. Finally, the research introduces a temporal analysis of learning behavior transitions, understanding of how student engagement evolves over time.