Machine Learning-Based Mental Health Risk Prediction and Behavioural Clustering among Malaysian Undergraduates

Authors

• Norsyela Muhammad Noor Mathivanan, School of Computing and Creative Media, University of Wollongong Malaysia, Shah Alam, Malaysia; UOW Malaysia KDU Penang University College, Penang, MalaysiaFollow
• Debbie Yong Xi Foo, School of Computing and Creative Media, University of Wollongong Malaysia, Shah Alam, MalaysiaFollow
• Gerard Soon Lee Chong, School of Computing and Creative Media, University of Wollongong Malaysia, Shah Alam, Malaysia; UOW Malaysia KDU Penang University College, Penang, MalaysiaFollow
• Siti Fazilah Shamsudin, School of Computing and Creative Media, University of Wollongong Malaysia, Shah Alam, Malaysia; UOW Malaysia KDU Penang University College, Penang, MalaysiaFollow
• Marizkays P. Jamison, College of Computing and Information Technologies, National University Philippines, Manila City, PhilippinesFollow

Corresponding Author

Norsyela Muhammad Noor Mathivanan ([email protected])

Abstract

Mental health issues among undergraduate students have become increasingly prevalent, affecting emotional well-being, academic performance, and overall quality of life. The growing number of students experiencing stress, anxiety, and depression highlights the need for effective predictive models to support early risk identification and personalized intervention strategies. This study investigates the performance of several machine learning approaches, including Logistic Regression (LR), Random Forest (RF), Gradient Boosting, LightGBM, Extra Trees, XGBoost, Support Vector Classifier (SVC), and a stacked ensemble model for predicting mental health risk among Malaysian undergraduate students. The study utilizes a dataset comprising demographic, academic, lifestyle, and psychological assessment attributes collected through structured self-assessment questionnaires. Model performance was evaluated using 10-fold cross-validation based on accuracy, precision, recall, and Macro F1-score metrics. Results indicate that Logistic Regression and Random Forest achieved superior predictive performance compared to other individual models and were subsequently integrated into a stacked ensemble framework to further improve classification capability. In addition, clustering analysis identified distinct behavioural characteristics among students with moderate and severe mental health risk levels. The findings highlight the potential of integrating predictive analytics and behavioural clustering in intelligent mental health support systems. This study contributes to the growing field of machine learning applications in mental health analytics and provides valuable insights for higher education institutions in supporting undergraduate well-being.

Publication Date

10-1-2026

Volume

11

Issue

2

Recommendation of Reviewers

yes

Recommended Citation

Mathivanan, Norsyela Muhammad Noor; Foo, Debbie Yong Xi; Chong, Gerard Soon Lee; Shamsudin, Siti Fazilah; and Jamison, Marizkays P. (2026) "Machine Learning-Based Mental Health Risk Prediction and Behavioural Clustering among Malaysian Undergraduates," Malaysian Journal of Computing (MJoC): Vol. 11: Iss. 2, Article 9.

Available at: https://mjoc.researchcommons.org/journal/vol11/iss2/9