From Clicks to Class Participation: Demystifying Student Engagement Factors with Attribute Ranking and Predictive Analytics

• Published in: 2024 5th International Conference on Artificial Intelligence and Data Sciences, AiDAS 2024 - Proceedings
• Main Author: Hanapi R.; Razali M.N.; Yahya S.; Seman S.A.A.; Halamy S.; Rahim E.A.
• Format: Conference paper
• Language: English
• Published: Institute of Electrical and Electronics Engineers Inc. 2024
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85209651309&doi=10.1109%2fAiDAS63860.2024.10730411&partnerID=40&md5=243b91357e34449bdf06afc35ae1de55

Student engagement is important in virtual and physical learning environments to ensure students are connected with the learning materials and the learning environment. However, understanding student engagement is difficult due to limited observation and uncertain participation metrics. This paper explores the application of machine learning techniques to analyze student engagement levels in online and face-to-face learning environments to uncover actionable insights. The data about student engagement was collected comprising diverse student attributes such as class interest, attendance, effort, and technology proficiency. Firstly, the dataset underwent preprocessing to handle missing values and normalize attributes. Attributes selection using Information Gain and Correlation are employed to identify key predictors influencing student engagement. Subsequently, classification algorithms were applied to build predictive models. Model performance is evaluated using metrics such as accuracy, precision, recall, and F1-score. Afterwards, trend analysis was carried out using a data visualization approach. The insights obtained from these analyses guide to the identification of critical factors such as class interest, attendance, and technology proficiency, which significantly impact engagement levels. The findings highlight the effectiveness of machine learning in discerning patterns and predicting student behaviors in virtual learning environments with Logistic Regression yielded 93.30% classification accuracy. Practical implications such as targeted interventions and curriculum adaptations can be designed to enhance engagement based on the identified predictors. © 2024 IEEE.