KMAN Publication Inc. (KMANPUB) International Journal of Innovation Management and Organizational Behavior (IJIMOB) 3041-8992 6 Serial Number 26 2026 05 01 Predicting Employee Engagement Through Extreme Gradient Boosting (XGBoost): An Explainable AI Approach Predicting Employee Engagement Through Extreme Gradient Boosting (XGBoost): An Explainable AI Approach 1 14 10.61838/kman.ijimob.5426 EN Journal Article 2025 11 07 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> This study aimed to predict employee engagement using Extreme Gradient Boosting (XGBoost) and Explainable Artificial Intelligence (XAI) techniques while identifying the relative importance of psychological, organizational, leadership, and demographic factors influencing employee engagement among employees in South African organizations.</p> <p><strong>Methods and Materials:</strong> This quantitative cross-sectional study was conducted among 1,248 employees working in diverse South African organizations across multiple industries. Data were collected using standardized instruments measuring employee engagement, psychological empowerment, perceived organizational support, job satisfaction, psychological safety, and transformational leadership, alongside demographic and organizational variables. Following data preprocessing, feature engineering, and missing value treatment, the dataset was divided into training and testing subsets using an 80:20 ratio. Extreme Gradient Boosting (XGBoost) served as the primary predictive model and was compared with Multiple Linear Regression, Decision Tree Regression, Support Vector Regression, and Random Forest models. Model performance was evaluated using coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). Explainable Artificial Intelligence was implemented using SHapley Additive exPlanations (SHAP) to determine feature importance and interpret model predictions.</p> <p><strong>Findings:</strong> The results demonstrated that XGBoost outperformed all competing models, achieving the highest predictive accuracy (R² = 0.902) and the lowest prediction errors (RMSE = 0.276, MAE = 0.198, and MAPE = 4.88%). Random Forest exhibited the second-highest predictive performance, while Multiple Linear Regression produced the weakest results. SHAP analysis revealed that psychological empowerment was the most influential predictor of employee engagement, followed by job satisfaction, perceived organizational support, transformational leadership, and psychological safety. Additional contributors included performance ratings, training participation, organizational tenure, and workload balance. The explainability analysis further indicated that higher levels of these organizational and psychological resources consistently generated positive effects on engagement predictions, whereas demographic characteristics exhibited comparatively limited predictive influence.</p> <p><strong>Conclusion:</strong> The findings demonstrate that employee engagement can be predicted with high accuracy using XGBoost and explainable artificial intelligence techniques. Psychological empowerment, job satisfaction, organizational support, transformational leadership, and psychological safety emerged as the most critical drivers of engagement. The integration of predictive analytics and explainable AI provides organizations with a powerful evidence-based framework for understanding, forecasting, and enhancing employee engagement while supporting strategic human resource decision-making.</p> </td> </tr> </tbody> </table> https://www.journals.kmanpub.com/index.php/ijimob/article/download/5426/10546