Exploring Incremental Classifiers For Predicting Student Performance Across Semesters

Accurately predicting student performance in advance has become crucial in educational systems to ensure timely interventions and enhance academic support. This research investigates the application of incremental learning algorithms to predict student grades across multiple semesters for a course. The traditional classifiers may not suit the student’s academic prediction since the data is being added frequently, so there is a necessity of using classifiers that support the incremental data. Various incremental classifier categories, such as Naive Bayes, linear models, tree-based models, ensemble methods, and neighbor-based models, were used to assess their efficacy in real-time student data. Student performance data for four semesters are used to study the dynamic learning process of the classifier. Each semester’s data is trained and updated with the previous semester’s model. This allows each model to adapt when new data is introduced every semester. The study will examine how well these incremental models perform on incremental data by tracking performance measures such as accuracy, precision, recall, F1 score, and training time for each semester. The findings will highlight the suitable incremental classifier to predict the student’s grade over various periods and will provide key insights for educators.