Decoding Online Student Behavior and Procrastination Using Clustering and Predictive Analytics

In the age of online education, it is crucial to comprehend the behavioral elements that influence student performance in order to improve learning outcomes and facilitate personalized approaches. This study proposes a data-driven framework to model and predict student behavioral patterns associated with procrastination in online learning environments. Rather than directly predicting procrastination as a binary construct, we operationalize it as a latent behavioral spectrum using indicators derived from submission timing and engagement patterns. K-Means clustering is first applied to identify six distinct latent behavioral profiles. These clusters, interpreted as procrastination-related patterns (ranging from strategic delay to dysfunctional procrastination), are subsequently predicted using supervised learning models including random forest, XGBoost (XGB), support vector machines (SVM), and logistic regression. To ensure interpretability and algorithmic fairness, we integrate explainable artificial intelligence (AI) techniques using SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME). Experimental results demonstrate that the proposed framework effectively captures meaningful behavioral patterns with a peak predictive accuracy of 98.36% (SVM). Furthermore, the Explainable AI (xAI) analysis provides actionable, behaviorally-grounded insights, thus offering a robust foundation for early, equitable intervention in online learning environments and Smart Campuses.