A Systematic Review of the Machine Learning-Based Educational Data Mining Models for Student Performance Prediction

Abstract

Predicting students’ performance is one of the applications of Educational Data Mining (EDM). EDM uses big data from education setting to predict students’ performance .This provides an avenue to track the academic progress of all students by educators. Machine-learning-based models, such as artificial neural networks, decision trees, and support vector machines, and their hybrids are the most frequently used models to predict students’ performance. These models have ability to efficiently extract hidden and useful information from large educational datasets. Following PRISMA framework, the review conducts an exhaustive literature search, emphasizing on the machine learning models used in prediction of the student performance. Key findings highlight the efficacy of machine learning based model used in predicting students’ performance, and their major drawbacks. It was clear that these models use irrelevant features (noisy data), limited number of students’ features, some of which are not within the current context of a particular learning environment to predict. This systematic review provides insights into existing machine learning models used in student performance prediction, contributing to the field's advancement and providing guidance for researchers and practitione