Enhancing Predictive HR Analytics Using Principal Component Analysis and Extreme Gradient Boosting
Authors: Rambabu Pasumarthy, PVVS Eswar Rao, Suresh Kumar Samarla, Satya Srinivas Maddipati, M Chilaka Rao
DOI: 10.87349/JBUPT/281011
Page No: 1-7
Abstract
This research investigates the application of advanced machine learning methods in Human Resource Management (HRM), focusing on predicting employee status using Principal Component Analysis (PCA) and Extreme Gradient Boosting (XGBoost). The study leverages a real-world HR (Human Resource) dataset, where PCA was employed to reduce dimensionality and eliminate redundancy, thereby improving computational efficiency. XGBoost was then applied to develop a robust classification model capable of predicting employee status outcomes such as active, resigned, or terminated. The proposed methodology achieved strong performance, with overall accuracy exceeding 90% alongside balanced precision, recall, F1-score, and ROC-AUC metrics. The results highlight the importance of employee-related factors such as tenure, compensation, job satisfaction, and promotion history in shaping predictive outcomes. Moreover, the integration of PCA and XGBoost not only enhanced accuracy but also provided scalability for large organizational datasets. The findings demonstrate the potential of predictive HR analytics to support data-driven decision-making, optimize workforce planning, and reduce attrition risks while underscoring the need to address fairness and ethical considerations in AI-enabled HR systems.
