Academic course planning recommendation and students’ performance prediction multi-modal based on educational data mining techniques

Abstract

Educational Data Mining (EDM) has recently received significant attention, leading to the development of various Data Mining (DM) methodologies for extracting hidden knowledge within educational data. This knowledge is crucial for enhancing teaching methods and improving student learning experiences, ultimately contributing to better student performance and overall educational outcomes. Students confront difficulties in selecting appropriate courses and suitable departments, which is regarded as the most important factor in avoiding career failure. Predicting students’ academic performance is vital for evaluating the success of educational institutions. In this study, eleven Machine Learning (ML) algorithms and three Deep Learning (DL) algorithms namely Support Vector Classification (SVC), K-Nearest Neighbor (KNN), Logistic regression (LR), Decision tree (DT), Linear discriminant analysis (LDA), Quadratic Discriminant Analysis (QDA), Random Forest (RF), Gradient Boosting (GB), Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (Light GBM), Extra Trees, Deep Artificial Neural Network (DANN), Gated Recurrent Unit (GRU), and Long Short-Term Memory (LSTM), were evaluated using real dataset from the Faculty of Computers and Information Sciences (FCIS) at Mansoura University (MU). A prediction model was developed to predict students’ academic grades in upcoming courses based on their past performance, alongside a recommendation model for guiding students towards suitable courses and departments. The results demonstrate that the Support Vector Classification (SVC) model outperformed others, achieving a 78.04% multi-classification accuracy and a 75.37% F1-Score. This study underscores the potential of individual ML and DL models to predict students’ academic performance based on real dataset features.