LEARNING ANAYLYTICS PREDICTIVE MODELING IN E-LEARNING PLATFORMS FOR EARLY IDENTIFICATION OF AT-RISK STUDENTS

Abstract

Early identification of at-risk students in e-learning platforms is often treated as a prediction problem, yet many studies do not carry this framing through to actionable early-warning decisions. A recurring gap is that risk indicators,
predictive modeling, timing, interpretation, and intervention are seldom connected into a coherent decision logic that clarifies what can and cannot be concluded. This paper therefore presents a conceptual framework that defines at-risk and early, organizes learning management system signals,
and situates modeling within a timing and interpretability layer that links to a response pathway. The framework is intended for literature synthesis
and conceptual analysis rather than empirical benchmarking, and it makes explicit the boundary between cautious early-warning statements and
stronger causal claims. Taken together, the contribution is a disciplined
template for reporting predictive learning analytics in terms of practical usefulness, fairness, and intervention readiness, with attention to e-learning settings relevant to learning analytics researchers and online student success practitioners.