Abstract:
This study explores the relationship between smoking, alcohol consumption, and psychological wellness among university students, employing machinelearning techniques to uncover patterns and provide predictive insights. Adataset of 1163 responses was analyzed, incorporating demographic, behavioral, and mental health-related features. The study applied six machine learningmodels Logistic Regression, SVM, KNN, XGBoost, Stacking Classifier, andanensemble of Deep Neural Networks to predict psychological wellness. Amongthese, the Stacking Classifier emerged as the most effective, achievinganaccuracy of 81%, showcasing the advantages of ensemble learning methods inhandling complex data patterns. The findings show a strong link betweendrinking and smoking and mental health outcomes, emphasizing the necessityof focused treatments to break negative patterns and advance mental health. When developing evidence-based plans to enhance students' well-being, policymakers, healthcare professionals, and educational institutions mayall
benefit from these results. The research highlights sustainability, ethical
issues, and the significance of treating data responsibly. Additionally, it offersa starting point for further study to improve mental health predictionandintervention methods, such as utilizing cutting-edge machine learningtechniques, integrating longitudinal data, and investigating other behavioral
aspects. Students' better lives and behavioral health analytics are advancedbythis research.
