This paper presents an artificial intelligence-based model for forecasting and maximizing student performance, utilizing intelligent classrooms that leverage real-time data collected through IoT and edge computing. Based on a dataset that reflects engagement measures, environmental characteristics, and response times, we trained a series of machine learning models to categorize student performance. Logistic Regression was the most successful model, with an accuracy of 95%, a precision of 95.3%, a recall of 95%, and an AUC of 1.000, which outperforms Random Forest and Gradient Boosting. The SHAP analysis provided that Learning Outcome, Attention Level, and Activity Type were the most significant characteristics. The proposed application features an interactive dashboard that enables teachers to input information on classroom conditions and receive performance predictions, along with corresponding explanations. The study makes contributions to adaptive learning and innovative learning feedback systems.