An early warning system for students at risk using supervised machine learning

• Main Authors: Yam, Zheng Hong, Mohd Norshahriel, Abd Rani, Nabilah Filzah, Mohd Radzuan, Lim, Huay Yen, Sarasvathi, Nagalingam
• Format: Article
• Language: English
• Published: School of Engineering, Taylor’s University 2024
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://umpir.ump.edu.my/id/eprint/42192/1/An%20early%20warning%20system%20for%20students%20at%20risk%20using%20supervised%20machine%20learning.pdf; http://umpir.ump.edu.my/id/eprint/42192/; https://jestec.taylors.edu.my/Special%20Issue%20on%20ICIT2022_2/ICIT2022_2_12.pdf; https://jestec.taylors.edu.my/Special%20Issue%20on%20ICIT2022_2/ICIT2022_2_12.pdf

The Covid-19 pandemic has brought numerous social issues to the fore, including poverty alleviation and education. As a result, significant changes have occurred in education, in which teaching is done remotely. To keep up with the course's pace, students must be self-motivated, well-organized, and have time management skills. Without these behaviours, online education is inappropriate for students. According to the research, 52% of students who sign up for a course would never read the course materials. Furthermore, throughout the course of five years, the dropout rate reached a stunning 96%. The main objective of this study is to create an early warning system for educators to use in educational institutions. For that, this study will perform a study of the current issues, factors, and solutions in education student’s data, determine the supervised machine learning algorithms, compare which model is the best predict the students’ performances, develop an early warning system for the educators to make an early decision in order to assist and consult the at-risk students, and finally conduct testing and evaluation of the system. Besides, this study also focuses on the Decision Tree, Random Forest, and XGBoost models in the system. The system will detect or forecast symptoms of dropout or potential dangers ahead of time, allowing educational institutions to anticipate problems and provide adequate educational services through appropriate intervention and response. A dashboard system will be created with a descriptive data mining model that can examine and make early judgments on at-risk students before they become highrisk. The web-based platform enables educational institutions to explore data patterns and analyse current crucial key performance metrics conditions. The technology will display the analysed results as well as visualized data to help educators gain insight and make better decisions.