AI-Driven Student Performance Prediction Using Multi-Class Classifiers

The integration of Artificial Intelligence (AI) into education has significantly transformed traditional teaching and learning practices by enabling personalized and adaptive learning experiences designed to meet the unique needs of individual students. This study introduces a comprehensive AI-driven framework for predicting student performance using a variety of machine learning classifiers, including J48 Decision Tree, Random Forest, Naïve Bayes, Support Vector Machine (SVM), and K-Nearest Neighbors (KNN). The framework evaluates multiple dimensions of student data—academic achievements, behavioral patterns, psychological factors, and skill assessments—to categorize learners into three distinct groups: fast, moderate, and slow performers, thereby facilitating targeted and timely academic interventions. The research utilized a dataset comprising 100 student records with 39 attributes, which underwent extensive preprocessing steps such as normalization, feature selection, and data anonymization to maintain privacy and ensure consistency. Experimental analysis demonstrated that the J48 Decision Tree classifier outperformed the others, achieving 100% accuracy, followed by Random Forest (98.5%) and SVM (96.7%). Further correlation analysis revealed that attributes such as aptitude, attendance, and motivation had a strong positive impact on CGPA, while stress levels showed a negative relationship. Additionally, an ANOVA test confirmed the statistical significance of these findings, validating the robustness of the proposed model. This framework highlights the potential of AI to revolutionize education by supporting real-time, personalized interventions and empowering educators with actionable insights for decision-making. Future work will explore the integration of deep learning models, real-time feedback mechanisms, and scalability to accommodate diverse educational settings.