Research on Student Performance Prediction Based on Deep Learning and Multimodal Data Fusion

This study proposes a student performance prediction framework based on the XGBoost and LSTM integrated model, called the XGLNet Framework. By combining nonlinear features and time series information, it enhances the accuracy of student performance predictions. The research utilizes student learning data, applying the XGLNet Framework for analysis and prediction. Experimental results show that the XGLNet Framework performs excellently across several key performance metrics, including Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and the Coefficient of Determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>). Compared to standalone XGBoost and LSTM models, the XGLNet Framework significantly reduces prediction errors and improves model fit. It effectively captures both nonlinear changes and temporal trends in student performance. Future research can further optimize the framework to enhance its applicability and robustness in broader educational scenarios.