Adaptive Mathematics Lesson Planning for Personalized Learning

Mathematics is crucial in learning core sciences and resolving everyday problems. However, conventional teaching methods frequently cannot cater to the specific needs of learners, especially in subjects with high complexity levels like mathematics, leading to disengagement and poor performance. This study introduces a machine learning-based adaptive learning system that individualizes learning mathematics using four fundamental components. Collaborative learning is enhanced through hierarchical math lesson preference and peer study prediction, with XGBoost being 88.4% accurate on lesson preference and 85.7% accurate on peer prediction. Cognitive behavior analysis is conducted through gamification with Self-Organizing Maps (SOMs) clustering students based on cognitive traits, effectively reducing quantization error under 10,000 iterations. Support Vector Machines (SVMs) are 81.4% accurate for cognition level prediction. Stress detection, using voice and facial expression analysis, is 85.3% accurate, with the stress relieved through music recommendation based on K-Means. Finally, individualized lesson planning is being optimized using classification techniques and obtained over 90% accuracy from SVM. Future work involves real-world deployment and its impact on student achievement.