LEARNING ANALYTICS FOR ONLINE EDUCATION: Predicting Academic Performance and Enhancing Engagement through Data-Driven Models

This book investigates the pressing issues of learner engagement and academic attrition in online education environments. With a focus on technical learners in Karnataka, India, the research introduces the EDU Insight framework to analyze key behavioral and demographic factors impacting student performance. It proposes a score prediction model using random forest and synthetic data augmentation through tabular GANs to forecast learner outcomes with high accuracy. Additionally, a hybrid ensemble learning approach incorporating weighted classifiers and meta-learners is developed to further refine predictive performance. To support personalized learning, an autoencoder-based collaborative filtering recommendation system is introduced, tailoring course suggestions based on learner behavior and demographics. The study's integrated use of learning analytics and machine learning contributes novel methodologies for predictive accuracy, data privacy, and personalized learning interventions in online education systems.