KMAN Publication Inc. (KMANPUB) International Journal of Innovation Management and Organizational Behavior (IJIMOB) 3041-8992 6 Serial Number 27 2026 07 01 Explainable XGBoost Prediction of Employee Turnover Intention Using Organizational Behavior and Human Resource Features Explainable XGBoost Prediction of Employee Turnover Intention Using Organizational Behavior and Human Resource Features 1 17 EN Journal Article 2026 04 19 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> This study aimed to develop and interpret an XGBoost-based machine-learning model for predicting employee turnover intention among employees in France using organizational behavior and human resource features.</p> <p><strong>Methods and Materials:</strong> This quantitative, cross-sectional predictive study included 684 employees working in public- and private-sector organizations in France. Participants completed a structured questionnaire assessing turnover intention together with organizational behavior and human resource variables, including job satisfaction, organizational commitment, perceived organizational support, work engagement, job autonomy, role clarity, role conflict, work-related stress, organizational justice, supervisor support, work–life balance, compensation satisfaction, career advancement opportunities, recognition and rewards, job security, leadership quality, perceived internal mobility, and HR fairness. The dataset was stratified into training and independent test subsets. An Extreme Gradient Boosting model was developed using stratified five-fold cross-validation and hyperparameter optimization. Predictive performance was evaluated using ROC-AUC, accuracy, precision, recall, specificity, F1 score, balanced accuracy, log loss, and Brier score. SHapley Additive exPlanations were applied to interpret global and individual feature contributions.</p> <p><strong>Findings:</strong> The final XGBoost model demonstrated strong discrimination in the independent test set, with an ROC-AUC of 0.88, accuracy of 0.83, precision of 0.78, recall of 0.76, specificity of 0.87, F1 score of 0.77, and balanced accuracy of 0.82. The test log loss and Brier score were 0.39 and 0.13, respectively. SHAP analysis identified job satisfaction as the most influential predictor, followed by organizational commitment, work-related stress, career advancement opportunities, perceived organizational support, work–life balance, compensation satisfaction, leadership quality, recognition and rewards, and perceived HR fairness. Lower satisfaction, weaker commitment and support, higher stress, and fewer career opportunities consistently increased predicted turnover intention.</p> <p><strong>Conclusion:</strong> Explainable XGBoost provided accurate and interpretable prediction of employee turnover intention, demonstrating that modifiable organizational and HR-related conditions were more influential than demographic characteristics and may therefore represent priority targets for evidence-based retention strategies.</p> </td> </tr> </tbody> </table> https://www.journals.kmanpub.com/index.php/ijimob/article/download/5682/11856