Delving Academic Performance and Stress Among Students: Insights Using Statistical and Machine Learning Methods

Context: Lifestyle factors have a significant impact on students' stress levels and academic performance. Mental health issues and academic demands are growing, it is essential to comprehend these connections for the purpose of early intervention and policy development. Aims: To determine how undergraduate students' lifestyle choices, stress levels, and academic achievement relate to one another. Additionally, the study used machine learning algorithms to create predictive models for stress level classification. Settings and Design: A cross-sectional analytical study Methods and Material: Carried out using secondary data gathered from 2,000 Indian undergraduate students via a structured online questionnaire. Statistical analysis used: The Kruskal-Wallis, Mann-Whitney U, ANOVA, correlation analyses, and descriptive statistics were used. Three levels of stress were distinguished. Support Vector Machine (SVM), Random Forest, XGBoost, K-Nearest Neighbors (KNN), and Logistic Regression were used for predictive modeling. Results: High levels of stress were reported by 51.5% of the students. While sleep and physical activity were negatively connected with stress levels, study hours were positively connected with GPA (r = 0.73).Statistically significant differences between stress categories and academic performance (p < 0.001) noted. Random Forest and XGBoost were the machine learning models with the highest predictive accuracy (100 percent), followed by SVM (95 percent) and KNN (90 percent). Conclusions: Students' stress levels and academic performance are greatly influenced by their lifestyle choices.