Article Dans Une Revue Asian Journal of Probability and Statistics Année : 2025

Predicting Burnout in College Students: A Machine Learning Approach Using Decision Tree and Psychometric Data

Résumé

Aims: This study aimed to model and predict academic burnout among college students using psychological and academic variables, and to identify the most influential factors contributing to burnout risk. Study Design: This research employed a quantitative, predictive research design utilizing decision tree analysis to classify students into burnout and non-burnout categories. Place and Duration of Study: The study was conducted at a private higher education institution in Misamis Occidental, Philippines, from January to March 2025. Methodology: A total of 291 college students participated through convenience sampling using an online survey. The instruments used included the Maslach Burnout Inventory (MBI), a tool designed to measure emotional exhaustion, cynicism, and academic efficacy along with the DASS-21 for stress, anxiety, and depression, the Brief COPE inventory, and a sleep quality measure. Descriptive statistics were used to profile respondents’ demographics and psychological traits. A Decision Tree model was applied to classify students as “burnout” (MBI score ≥ 3.0) or “not burnout,” and feature importance was analyzed to determine key predictors. Model performance was assessed using Recall, F1-score, and the Receiver Operating Characteristic–Area Under the Curve (ROC-AUC), which evaluates the model’s ability to distinguish between burnout and non-burnout cases. Results: Out of 291 respondents, 139 (47.8%) were classified as experiencing burnout, while 152 (52.2%) were not. Stress score (29.5%), coping strategies (26.4%), and sleep duration (15.4%) emerged as the top predictors. The model achieved a recall of 51.6%, F1 score of 59.3%, and ROC-AUC of 66.9%, indicating moderate predictive power. Descriptive analysis revealed that most respondents were male and reported moderate stress and good sleep quality, which may contribute to the lower burnout rates observed. Conclusion: Stress, coping strategies, and sleep behaviors are the most influential factors in predicting student burnout. It is recommended that future studies consider a female-majority sample and use more advanced machine learning models such as Random Forest to improve prediction accuracy and understanding of burnout patterns.

Fichier non déposé

Dates et versions

hal-05068212 , version 1 (15-05-2025)

Identifiants

Citer

Rachelle P Tapio. Predicting Burnout in College Students: A Machine Learning Approach Using Decision Tree and Psychometric Data. Asian Journal of Probability and Statistics, 2025, 27 (5), pp.50-60. ⟨10.9734/ajpas/2025/v27i5754⟩. ⟨hal-05068212⟩
31 Consultations
0 Téléchargements

Altmetric

Partager

  • More