The Use of Learning Analytics to Predict Student Success, Identify at-Risk Students, and Improve Academic Performance in Higher Institutions of Learning - A Systematic Literature Review.

Learning analytics has emerged as a powerful data-driven approach for enhancing student success and academic achievement in higher education. By leveraging educational data to monitor learner engagement, predict academic performance, and identify at-risk students, learning analytics has gained increasing attention among researchers and practitioners. However, evidence regarding its effectiveness, implementation challenges, and ethical implications remains dispersed across the literature. This systematic review synthesizes current evidence on how learning analytics supports student success and improves academic outcomes in higher education. I retrieved studies published between 2015 and 2025 from Google Scholar and screened them using predefined eligibility criteria. Of the 44 studies initially identified, 35 met the inclusion criteria after removing duplicates, studies published before 2015, and one retracted publication. I also used a narrative synthesis approach to analyze the findings. The review revealed that learning analytics primarily utilizes data from Learning Management Systems (LMS), assessment results, and student engagement indicators to predict academic performance and identify learners at risk of failure or withdrawal. Predictive models based on educational data mining and machine learning techniques enable timely interventions through early warning systems, personalized feedback, and targeted academic support, contributing to improved student retention, engagement, and achievement. Nevertheless, factors such as data quality, institutional context, and learner characteristics influence the effectiveness of these interventions. The review also identified significant concerns related to data privacy, informed consent, transparency, and algorithmic bias. Moreover, reliance on LMS-generated data alone may limit predictive accuracy by failing to account for cognitive, emotional, social, and socio-economic factors that influence student success. Overall, the findings suggest that learning analytics can strengthen evidence-based decision-making and enhance student outcomes when implemented within sound pedagogical and ethical frameworks. The review recommends integrating multidimensional data sources, developing context-sensitive predictive models, and establishing robust data governance mechanisms. Future research should investigate the long-term effects of learning analytics on academic achievement, retention, graduation rates, and educational equity across diverse higher education contexts.