AI-Based Occupational Stress Level Prediction Using Xgboost Algorithm

Abstract: In contemporary workplaces, occupational stress has become a major issue that has a detrimental impact on worker productivity, mental health, and organisational effectiveness. There is a need for clever and data-driven solutions because traditional stress assessment techniques are frequently subjective and not scalable. This work suggests a supervised machine learning-based artificial intelligence system for forecasting occupational stress levels. Data cleaning, encoding, and normalisation techniques were used to preprocess a structured dataset that included psychological, work-related, and demographic characteristics. In order to correctly classify stress levels and identify stress patterns in the dataset, the suggested model uses a classification algorithm. The XGBoost model outperformed baseline classifiers with 84% accuracy, 91% precision, 89% recall, 90% F1-score, and 0.91 ROC-AUC. The results show that early stress diagnosis and preventive measures in organisational settings can be successfully supported by AI-driven stress prediction systems. Keywords: Data analytics, machine learning, artificial intelligence, stress prediction, and classification