Authors: Assistant Professor Dr.J.Srilekha
Abstract: Attrition among employees is one of the greatest concerns in the modern organization due to high expenses caused by recruiting new personnel, productivity losses, and reduction of corporate knowledge. In the research under consideration, there is a developed predictive modeling system intended to predict employee attrition and develop appropriate retention strategies. The proposed methodology uses complex machine learning techniques such as Random Forest, Gradient Boosting, and ensembles along with mechanisms of explainable artificial intelligence that provide transparency of the model and facilitate its application. By applying the IBM HR Analytics Employee Attrition dataset containing information about employees' features, the framework uses advanced techniques of preprocessing, class balancing, and feature engineering to take into account complex aspects of attrition. As a result of experiments, it is revealed that the ensemble model demonstrates better forecasting results compared to baseline models. The interpretation of results based on the SHAP technique shows that overtime, monthly income, length of service, and job satisfaction are the most important factors affecting attrition. Thus, the research provides an innovative approach that will help turn human resource management into proactive workforce planning.
DOI: http://doi.org/
