Predicting User Engagement in E-Learning Platforms Using Decision Tree Classification: Analyzing Early Activity and Device Interaction Patterns

Abstract

This study investigates the prediction of user engagement in e-learning platforms by applying a Decision Tree classification model. Early user activity and device interaction patterns are explored as key predictors of engagement levels. With increasing demand for personalized learning strategies, identifying patterns of engagement early in the learning process can provide valuable insights for improving retention and learner outcomes. The dataset used in this study consists of various features, including user activity metrics (e.g., homework completion, task performance) and device interaction data (e.g., operating system, device type). After preprocessing and feature selection, a Decision Tree classifier was trained on the dataset to predict user engagement. The model's performance was evaluated using accuracy, precision, recall, and F1-score metrics. The results revealed that the Decision Tree model achieved an accuracy of 74.24%, with precision for the low-engagement class significantly lower than that for high-engagement users, indicating challenges in predicting less-engaged users. The study highlights the potential of using early engagement signals to predict learner behavior, providing a foundation for the development of personalized interventions. While the model provides useful insights, the study also acknowledges limitations, including dataset imbalance and limited generalizability across different e-learning platforms. Future research could explore the inclusion of additional engagement indicators, such as emotional response or interaction with course content, and the use of more advanced machine learning techniques. Overall, this research contributes to the growing body of knowledge on AI-driven user engagement prediction in e-learning, offering practical implications for improving student retention and learning outcomes.