Optimized Perception-Based Student Behavior Analysis Using Machine Learning

Students from diverse backgrounds face various academic challenges. Their performance depends on multiple factors, including knowledge level, language, culture, family pressure, social interactions, and demographic aspects. This study identifies key reasons for underperformance and high performance using survey data from 276 students. We compare eight classification techniques, where Logistic Regression achieves 94.5 % accuracy. Optimization with meta-heuristic algorithms, including Differential Evolution, further improves accuracy to 97.2%