Student Performance Prediction Using Machine Learning and Explainability

The increasing digitalization of academic activities has created large volumes of structured and unstructured student data. Traditional evaluation strategies rely heavily on retrospective assessments and fail to provide early intervention insights. This paper presents a machine-learning-based Student Performance Prediction System using Linear Regression and Explainable AI (SHAP) to forecast a student's final CGPA. The model integrates academic history, attendance, quiz performance, test averages, and assignment behavior using a sector-weighted feature engineering technique. A Streamlit-based interface enables individual and batch predictions while producing transparent explanation dashboards. Experimental results demonstrate a Mean Absolute Error (MAE) of 1.17 and RMSE of 1.37, with Linear Regression outperforming Random Forest and XGBoost. SHAP visualizations enhance interpretability and support data-driven academic decision-making. The system effectively identifies “At-Risk� students and generates intervention-ready reports for educational institutions.