A Probabilistic Bayesian Network Framework for Strategic Human Resource Retention and Employee Attrition Risk Modeling

Le résumé fourni par la source

Employee attrition creates several challenges for organizations by causing them to lose their staff's skills, spend more on recruiting new staff and experience breaks in regular operations. The typical traditional approach to attrition prediction prizes accuracy of classification, but it is not interpretable, flexible or able to handle uncertainties which are necessary for good HR decisions. The study will build a predictive and understandable model using Bayesian Networks (BN) to help anticipate employee turnover and back up company actions in human resources. The new aspect of this study comes from using BN modeling which allows the researchers to estimate attrition rates and also highlight how key variables in HR like job satisfaction, monthly income, work-life balance and job role are related to each other. The suggested BN framework allows HR experts to test out different scenario ideas and imagine how the results of their actions might change. The BN model was trained on the IBM HR Analytics Employee Attrition and Performance dataset and validated, providing accuracy of 98.3%, precision of 98.5%, recall of 92.4% and F1 score of 95.3% which is higher than the average accuracy found in both logistic regression and random forest. This framework helps move the field of HR analytics forward by connecting promising trends with useful and easy-to-understand outcomes.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.