Workplace Stress Detection and Mental Health Prediction Using Machine Learning

Workplace stress is a significant factor contributing to health, job satisfaction, and performance at work. A lengthy duration of stress exposure is harmful to mental and physical health. Traditionally, stress was assessed by means of the self-reported survey data and human analysis which can produce subjectivity. This paper examines a new machine-learning based detection of workplace stress using survey data. Statistical analysis will be utilized to determine how much statistically each question factor will weigh in to the model, ensuring the most taxing factors are weighted most heavily when modelling. Following this, the responses of the weighted questions will be used in XGBoost Classifier and the results indicate a of 99.4% meaning workplace stress can be detected reliably and automatically without human feedback. In addition, an equation was included in order to compute a weighted stress score that increases interpretation of the data. Finally, this study demonstrate the benefits of a data-based approach to stress detection and also how organizations legally have the tools to track and measure well-being.