Data-Driven Analysis of Learning Behaviors Among At-Risk Students Across Disciplines Using Data Mining Techniques

In recent years, the proliferation of e-learning systems has provided an unprecedented opportunity to collect and analyze vast amounts of data concerning student learning behaviors. This research aims to use data mining techniques to identify common learning patterns of at-risk students across various disciplines. Unlike previous studies that have predominantly focused on individual disciplines or subject areas, this study integrates data from multiple disciplines to offer a comprehensive analysis.

We utilized big data sourced from log files of an online e-learning system called Blackboard, supplemented by performance data from individual assessment components. The dataset included all students from 240 subjects across four academic divisions, resulting in a total of 1.7 million rows of records. Our findings show that at-risk students often exhibit low click rates on the e-learning system, delay starting assignments until close to deadlines, and consistently low participation in online activities. Additionally, the study found a high correlation between students' performances in in-class exercises and tests with their overall subject performance. This suggests that regular engagement and participation in formative assessments are strong indicators of academic success.

These insights are critical as they will enable educators to develop targeted interventions aimed at improving student performance and retention rates. By understanding the distinctive learning behaviors of at-risk students, educators can provide timely support and resources to help these students succeed. This research not only highlights the importance of using data-driven approaches in education but also underscores the potential of interdisciplinary analysis in enhancing educational outcomes.