Learning analytic framework for students’ academic performance and critical learning pathways

• Main Authors: Lyn, Jessica Tan Yen, Goh, Yong Kheng, Lai, An Chow, Ngeow, Yoke Meng
• Format: Article
• Language: English
• Published: Penerbit Universiti Kebangsaan Malaysia 2024
• Online Access: http://journalarticle.ukm.my/24109/1/127_147%20Paper_10.pdf; http://journalarticle.ukm.my/24109/; http://www.ukm.my/jqma

In the domain of higher education, the need to leverage data-driven insights for understanding and enhancing student academic performance is becoming increasingly critical. To address this, a unified learning analytics framework is proposed, aimed at deciphering complex student academic journeys and fostering data-informed decision-making for educational institutions. This framework’s methodology involves several key steps, starting with standardized data collection and pre-processing. Subsequently, dimensionality reduction techniques like Principal Component Analysis (PCA) and Non-negative matrix factorization (NMF) are applied to capture the most influential course components and grade information. The resulting reduced dataset is then subjected to various clustering algorithms, including partition-based clustering (K-means), hierarchical clustering, and density-based clustering (DBSCAN). These algorithms group students based on academic performance and course profiles, facilitating the identification of clusters with similar characteristics and academic trajectories. Furthermore, a collective network graph is constructed to analyze course relationships and program pathways, identify critical courses, and reveal influential factors affecting student performance and outcomes. This network analysis enables educators to identify bottleneck courses and areas that may require additional support or improvement, fostering a data-driven approach to curriculum design and enhancement. To showcase the framework’s efficacy, a case study was conducted on 3550 undergraduates from an engineering program at a Malaysian private university. The student dataset used in this study spans from 2005 to 2021, covering a wide range of academic years for analysis. The results demonstrate the framework’s capability to unveil valuable insights into students’ academic journeys, revealing key factors contributing to their success. By providing a holistic perspective of student performance and course interactions, the proposed learning analytics framework holds great promise for educational institutions seeking data-driven strategies to enhance student outcomes and optimize learning experiences.