Leveraging Machine Learning and AI to Enhance Educational Learning Analytics

Abstract

Data-driven insights play a pivotal role in optimising learning analytics within higher education institutions. Despite their importance, much of the data in these institutions remains untapped, trapped in siloed data stores. This study addresses this challenge by applying machine learning and mathematical modelling using a learning analytics research framework (Khalil et al., 2022), encompassing phases of data insights, analytics, and intervention. The study aimed to identify features influencing learner progression and discover student groups in educational environments. Utilising data from 1,017 students over a 3-year period, the research employed unsupervised machine learning techniques and automated student feedback. Feature analysis identified attendance, interactions, time intervals, and activities related to quizzes and workshops as useful predictors of students requiring additional support. The study found that K-means model performs best with an average recall of 89% and an overall accuracy of 72% for identifying at-risk student groups using non-personally identifying data. The findings emphasise the utility of unsupervised machine learning for early identification of at-risk students, enabling timely interventions to prevent potential failure or dropout. Personalised and automated feedback forms summarising students’ progression received high ratings, 93% rating for usefulness from students, highlighting their satisfaction with it as a learning analytics intervention.