Predicting Student Academic Performance in the Virtual Learning Environment Based on Adaptive Weighted Ensemble Learning and Dynamic Error Correction

Abstract

This study presents an academic performance prediction framework in the virtual learning environment based on adaptive weighted ensemble learning and dynamic error correction. The framework first constructs static and dynamic features of students based on data extracted from the virtual learning environment. Then, extreme gradient boosting, light gradient boosting machine, random forest, and support vector machine are adopted as base models to predict the initial category probabilities of student academic performance. Furthermore, by learning an error feature vector consisting of seven indicators, an error correction model and learnable weighting are developed to correct the initial outputs of the base models. Finally, the framework uses an adaptive weighting mechanism to fuse the corrected outputs of the base models and produce the final predictions. This study conducts validation using the Open University Learning Analytics Dataset to demonstrate the superiority of the framework.