A multi classifier framework for academic performance prediction of students

Abstract

Predicting student academic performance is a critical research area in Higher Education Institutions (HEIs) that aims to enhance student success and improve educational outcomes. The present research study explores Educational Data Mining (EDM) and Machine Learning (ML) techniques to develop a predictive model to identify low-performing students as well as students at-risk of failure. It also facilitates educators to take proactive steps to improve the academic performance of students and assist the administration in making constructive decisions to enhance the teaching-learning process. Developing an accurate and fast student performance prediction system is an immediate need, and ML classifiers are prominently used for this purpose. ML classifiers that can scale effectively with the increase in feature space dimensions are highly desirable, therefore motivating the present research. The work presents a multi-classifier framework for academic performance prediction of students that can help in enhancing their academic performance by leveraging the student academic data available in HEIs. The proposed framework addresses limitations in existing research, which often focuses on a single classifier-based predictive model. The study emphasizes accurate student performance prediction by utilizing a dataset of PG students from Panjab University, Chandigarh, collected over multiple academic years. Focusing on key academic features such as grades from high school, undergraduate education, entrance exams, and semester results, the research demonstrates how existing academic data can be harnessed to generate insightful predictions. newline