Design and Development of Online College English Teaching System Based on Explainable Machine Learning

As online education becomes increasingly prevalent, the need for intelligent and interpretable teaching support systems has grown significantly. This research proposes the development of an Online College English Teaching System leveraging Explainable Machine Learning (XML) techniques to enhance teaching quality, student engagement, and personalized learning outcomes. The system integrates student behaviour data, quiz performance, participation logs, and linguistic feedback to train predictive models that assess comprehension, attention span, and learning progress. A variety of models including XGBoost, SHAP-enhanced Random Forests, and LIME-explained neural networks were evaluated using a benchmark dataset collected from an online English learning platform involving over 2,000 students. The best-performing model achieved a prediction accuracy of 98.6 for learning outcomes with interpretable insights into feature contributions. Explainability tools revealed that active participation, timely submission of assignments, and vocabulary test performance were the top three indicators influencing overall progress. This system not only ensures high prediction performance but also provides transparent visual explanations to educators, enabling data-driven decision-making in content adaptation and student support strategies. The integration of explainable AI advances the trustworthiness and effectiveness of intelligent teaching platforms in higher education.