Abstract
Teacher burnout is a persistent global challenge with significant consequences for educator wellbeing, instructional quality, and school climate. Despite extensive research, most studies rely on small local samples, predefined burnout scales, and limited analytical techniques, leaving gaps in understanding the latent structure of burnout and the factors that predict it across diverse educational systems. This study addresses these gaps by applying a hybrid machine learning framework to the OECD TALIS 2018 teacher dataset (N = 38,081) to discover latent burnout profiles and build predictive models capable of identifying teachers at risk. Unsupervised k-means clustering was used to uncover naturally occurring burnout groups based on stress exposure, workload demands, wellbeing indicators, and job satisfaction metrics. Two robust profiles emerged, High Burnout and Low Burnout, demonstrating clear separation in stress load, satisfaction levels, and composite burnout severity. These profiles were subsequently analysed across demographic, institutional, and country-level characteristics. Supervised machine learning modelswere then trained to predict profile membership using demographic, workload, wellbeing, and institutional variables. Neural networks and elastic-net logistic regression achieved the highest performance (Accuracy> 0.98, ROC–AUC> 0.99), with random forests and decision trees also performing strongly. SHAP analyses revealed that health impacts, stress from instructional and administrative tasks, and job satisfaction indicators were the most influential predictors. The findings provide cross-cultural evidence that teacher burnout forms distinct latent patterns that can be accurately predicted using multifactor data. The hybrid clustering prediction framework offers a scalable foundation for early warning systems and highlights actionable psychological and institutional factors that can guide targeted interventions to support teacher wellbeing.
Publication Date
4-2026
Document Type
Thesis
Student Type
Graduate
Degree Name
Professional Studies (MS)
Department, Program, or Center
Graduate Programs & Research
Advisor
Ayman Ibrahim
Recommended Citation
Alkhzaimi, Fatma Fraishan Abdulla Hassan, "Predicting Teacher Burnout Across Cultures: A Machine Learning Approach Using TALIS 2018 Data" (2026). Thesis. Rochester Institute of Technology. Accessed from
https://repository.rit.edu/theses/12581
Campus
RIT Dubai
Comments
This thesis has been embargoed. The full-text will be available on or around 5/3/2027.