Authors : Khandakar Rabbi Ahmed, Md Eahia Ansari, Md Naimul Ahsan, Towfika Salam, Rakibul Islam, Robiul Hoque, Nusrat Ameri
Publication date : 2025/10/23
Conference : 2025 IEEE 2nd International Conference on Computing, Applications and Systems (COMPAS)
Pages : 1-6
Publisher : IEEE
Description : Accurate and readable human resource (HR) projection is crucial for strategic personnel planning, especially in talent recruitment and turnover. The study presents HR-SelectRetain-XGDL, a combined machine learning model that merges Gradient Boosting (XGBoost), Deep Learning (DL), and Recursive Feature Elimination (RFE) to address the dual task of candidate selection and attrition prediction. When utilizing the IBM HR Analytics dataset, our model performs better when run through hierarchical feature selection, deep neural networks, and ensemble classification. The accuracy of 98.41% for candidate selection and 97.00% for retention prediction by the system developed is both better than two state-of-the-art baselines, DNN-RF and RF-RFE, on all the most essential measures of precision, recall, F1-score, and AUC. Systematic evaluations such as confusion matrices, ROC curve analysis, and training …
Scholar articles :
KR Ahmed, ME Ansari, MN Ahsan, T Salam, R Islam… – 2025 IEEE 2nd International Conference on Computing …, 2025