Computational Framework for Modelling Student Engagement in a University’s Team-Based Learning Ecosystem

Abstract

This chapter presents a learning analytics framework aimed at discovering salient factors that influence learning outcomes in a self-directed team-based learning (SDL-TBL) environment. The data used in this study consists of online logs and formative assessment scores from Year 1 and Year 2 curricula across two cohorts of students at the Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore. Firstly, descriptive analytics was performed on the frequency of online access to learning materials and individual readiness assessment (iRA) scores, independently, to compare the distribution of engagement features and iRA scores of the two cohorts. Secondly, to find the significant factors influencing learning, a predictive analytics layer was built using the engagement features to predict the learner’s iRA scores. Next, regression analysis was performed using boosted decision trees, both at the module and lesson-level to gain insights into factors of learner’s engagement that could influence their performance. Independent models were then built to predict aggregated mean iRA scores per module and iRA scores at each TBL lesson, using long-term and short-term engagement features, respectively. From the analyses, it is observed that short-term learning outcomes are influenced by engagement with media-based materials, whereas long-term learning outcomes are influenced by engagement in terms of downloads. Additionally, cumulative and consistent engagement were found to emerge as better predictors than promptness in engagement.