Predictors of emotional well-being in university professors using Machine Learning.

Background: In the current context of increasing psycho-emotional strain within higher education, the emotional well-being of university faculty has become a strategic variable for institutional management. In response to this challenge, the present study aimed to predict emotional well-being among faculty members from the State University of Milagro and the Technical University of Manabí through the application of machine learning algorithms. Methodology: A quantitative, explanatory, and correlational-predictive design was adopted, using a stratified probabilistic sample of 1,470 university professors. Data were collected through a psychometrically validated questionnaire and analyzed using supervised learning models implemented in Orange Data Mining. Results: The findings revealed that Gradient Boosting, Random Forest, and Neural Network algorithms achieved the highest predictive performance, reaching optimal levels of accuracy, sensitivity, and calibration. The most significant predictors identified were emotional regulation, social competencies, emotional autonomy, and emotional awareness, whereas life and well-being competencies did not show a positive relationship. Additionally, age and level of academic training were associated with higher levels of emotional well-being. Conclusion: The results highlight the capacity of machine learning algorithms to predict faculty emotional well-being with high accuracy and underscore their usefulness as decision-support tools for institutional management in occupational mental health.