Enhanced Student Performance Prediction Using XGBoost, LSTM, and Ensemble Learning Techniques

Student Performance Prediction System proposed in the study uses machine learning to predict with extreme accuracy the academic result of students. Feature engineering and advanced ML methods including XGBoost and LSTM have been employed to enhance the model accuracy. The system uses student academic records as a set of input features including grades achieved, attendance, test scores, and other relevant academic criteria. A combination of SHAP and LIME is used to ensure model interpretability and transparency. These methods help teachers in data-driven decision-making by showing which factors are most effective in influencing student performance. The performance of the model was evaluated numerically and its respective F1-score and Mean Absolute Error stand at 0.92 and 0.015, respectively. Findings from this indicate that students are labeled properly into performance categories. Such identification allows teachers and schools to improve educational outcomes for students through early intervention in the cases of difficult students through personalized learning strategies. The concept triggers a proactive mechanism in the management of Student performance where insights can be acted upon in real time. Behavioral data like studying habits, interest in figuring things, and extracurricular activities can be built into the system for further enhancement. Real-time monitoring features can be added to enable such projections to be modified on the fly, ensuring that therapies remain appropriate and effective over time.