International Conference on Reinventing Business Practices, Startups and Sustainability (ICRBSS) 2025

Employee attrition is a serious challenge for organizations in the stability, productivity, and financial
performance. The study aims to effectively harness predictive analytics and machine learning to predict employee turnover while determining key driving engagement factors for this outcome. It also deploys prediction models for attrition patterns using IBM HR Analytics Dataset through logistic regression, random forest, and support vector machine (SVM) models. Out of these models, the one with the best performance was found to be the random forest
model in predicting job turnover indicators, with an accuracy of 86.05% while actual work-related factors like job satisfaction, organization commitment, and career development also considered as forms of turnover indicators. By this measure, employee engagement comes out as the very important influencer of retention, where supervisor
involvement, work-life balance, as well as recognition are significant in reducing employee turnover. Further, it is found that new employees are quit-prone on average when there is less tenure and age, while higher salaries and better job security have lower attrition rates. Considering this perspective, this study posits that effective HR must take the charge of predictive analytics in the organization's approach towards retention improvement. Thus, data-driven insights can be used and followed by engagement actions for optimizing workforce management and mitigating attrition risks. This goes without saying that future research can revolve around strengthening the predictive models using advanced resampling techniques and qualitative insights that would allow understanding at a holistic level employee turnover.