Students Performance Prediction In Online Courses Using Machine Learning Algorithms

Abstract

The education system is changing fast with all the new technology. This means that schools now have a lot of information about how studentsre doing in school and how they behave. The old way of checking how well students are doing like giving them tests and homework takes a time and does not always find students who are struggling early enough. This paper is about a system that uses machine learning to figure out how well engineering students will do in school by looking at lots of things about them including how well they do, in their classes and how they behave. The new system uses ways to figure out how students are doing like Random Forest, bagging and boosting algorithms. It helps to put student outcomes into groups like pass or fail, good, okay or bad or many other groups. We compare these methods to see which one is the best at predicting how students will do and which one works the best. The system automatically finds students who are struggling so we can help them sooner. Give them the support they need to learn better. The student outcomes are classified into three-class and multi-class performance categories using these methods and the system is good at detecting, at-risk students. Additionally, the framework is designed to be scalable for large datasets and adaptable to diverse student populations. By integrating predictive analytics with a user-friendly interface and visualization tools, the proposed approach enhances decision-making for educators and institutional administrators. The experimental results demonstrate that ensemble methods outperform traditional classifiers in terms of accuracy and reliability, making the system a practical solution for modern educational performance evaluation.