Virtual Learning Environment to Predict Students' Academic Performance Using Deep Learning Technique

Data mining in education has gained importance in recent decades due to the abundance of educational data available. The purpose of this study is to investigate deep Learning and machine Learning methods for predicting students' academic performance. The study employed machine learning techniques, such as the decision tree, random forest, and logistic regression, and deep learning techniques, such as Deep Learning Long Short-Term Memory (LSTM). And it found that the LSTM model was best for sequential classification of time series and used the Open University Learning Analytics Dataset (OULAD) to develop predictive models. The LSTM achieved the highest accuracy, with an impressive 94.4%. This performance outperformed the decision tree, random forest, and logistic regression, models. The exceptional accuracy of the LSTM suggests that it can be a valuable tool for identifying at-risk students early and targeting interventions to improve their academic performance. This study recommends using this model in universities and colleges to predict the students' performance, and the prediction results would help improve the policymakers and decision-making.