Machine Learning Approach to Predict Student Performance Based on Previous Records

Educational Data Mining (EDM) is an emerging field dedicated to discovering and analyzing meaningful patterns in educational datasets. This paper provides a comparative analysis of seven machine-learning classifiers: Support Vector Machine (SVM), Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), K-Nearest Neighbors (KNN), Naive Bayes (NB), and Neural Network (NN). The primary objective is to evaluate these algorithms, combined with preprocessing techniques, for predicting student performance and identifying the most effective classifier using metrics such as the confusion matrix, accuracy, precision, and sensitivity. The dataset utilized in this study was obtained from Sebha University, Libya, comprising 6,000 student records. It includes 6 attributes from the university's SIS system and 43 attributes derived from completed questionnaires, focusing exclusively on students from scientific departments. The experimental results revealed that the RF algorithm outperformed other classifiers in predicting student performance. Significant improvements in the model's performance were observed after applying data preprocessing and feature selection techniques. These findings offer valuable insights to help new students select suitable academic programs, potentially enhancing their academic performance. Furthermore, the proposed model could serve as an early warning system to identify students at risk of academic failure, enabling timely intervention.