An integrated explainable artificial intelligence framework for employee attrition prediction and retention strategy generation

Employee attrition is a constant issue in competitive job markets. While predictive models can estimate attrition risk, turning those predictions into effective retention actions still poses challenges. This work aims to create a framework that links attrition prediction, explainable artificial intelligence (XAI), and generative AI to support data-driven and personalized retention strategies. Using the IBM HR Analytics Attrition dataset, we built a machine learning model to predict employee attrition. We applied SHAP explainability techniques to identify the main factors affecting individual attrition risk. We introduced an Employee Value Scoring (EVS) system to highlight high-value employees at risk. To translate insights into action, we used generative AI (Gemini) to create personalized retention recommendations based on the most important SHAP-derived features. The framework successfully identified high-risk employees and offered targeted, easy-to-understand recommendations based on individual attrition drivers. The results show how combining predictive modeling, explainability, and generative AI can help HR teams move from predicting risk to taking meaningful action. This work presents a new, unified approach that connects attrition prediction and effective retention planning. By integrating machine learning, XAI, and generative AI, the framework provides personalized and context-specific recommendations, improving the practical use of HR analytics for proactive talent management.