Enhancing Student Performance Prediction using E-Learning through Multimodal Data Integration and Machine Learning Techniques
NS Koti Mani Kumar Tirumanadham, S. Thaiyalnayaki, V. Ganesan
Abstract
Advanced machine learning techniques have shown significant promise in predicting student performance in E-Iearning systems, but challenges such as handling imbalanced datasets, integrating multimodal behavioral data, and managing the complexities of feature selection persist. Recent techniques, such as SMOTE for class balancing and BorutaNetCV for feature selection, have made strides in improving prediction accuracy; however, they often fail to address real-time cognitive state integration and the influence of diverse engagement patterns. To overcome these limitations, this study employs a comprehensive methodology that combines robust preprocessing methods, advanced feature selection, and ensemble machine learning models such as Random Forest, AdaBoost, and XGBoost. The findings demonstrate that predictive accuracy can be significantly enhanced through these techniques, and the proposed framework enables the development of personalized interventions that can foster improved student engagement and success in dynamic E-learning environments.