<![CDATA[Employee turnover remains a strategic challenge in human resource management because it affects workforce stability, recruitment costs, productivity, and long-term organizational sustainability. This study aims to analyze employee turnover risk and translate predictive analytics results into data-driven human resource retention strategies. A quantitative predictive analytics approach was applied using the IBM HR Analytics Employee Attrition and Performance dataset, consisting of 1,470 employee records. Data were processed using RapidMiner through attribute selection, nominal-to-binominal transformation, role assignment, nominal-to-numerical conversion, 10-fold cross-validation, and model evaluation. Four models were examined: Naive Bayes, Decision Tree, Random Forest, and Gradient Boosted Trees. The evaluation used accuracy, precision, recall, F1-score, AUC, and confusion matrix, with emphasis on the Attrition = Yes class as the main indicator of turnover risk. The results show that Gradient Boosted Trees provided the most balanced performance, achieving 84.49% accuracy, 52.04% precision, 48.52% recall, 50.22% F1-score, and 0.796 AUC. These findings indicate that predictive analytics can support early identification of employees requiring targeted retention interventions. This study contributes to HR analytics literature by positioning predictive modeling as a managerial decision-support tool for retention planning. Future research should use actual organizational data, longitudinal designs, and explainable AI methods to improve model interpretability and application.]]>
Copyrights © 2026
