Employee Salary Prediction in HRMS Using XGBoost with LLM and RAG

Read full text Summarise with VIRAT
Authors: R. Sravani, K. Renna
Cited by: 0
References: 2

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Accurate Salary Prediction is one of the key features of modern Human Resource Management Systems (HRMS). However, most traditional approaches often have limited ability to model the complex relationships between an employee’s attributes and the resulting salary. This work aims to develop an efficient Machine Learning-based salary prediction system utilizing XGBoost regression algorithm along with a Large Language Model (LLM) and Retrieval-Augmented Generation (RAG) for producing final salary prediction output that is human-readable and explainable, giving insights about factors affecting the final prediction result. In this research, the XGBoost model utilized different attributes including employee’s age, level of education, job title and years of experience to produce predicted salaries. Moreover, the RAG model enables retrieval of relevant data from the HR database, i.e. the organizational salary tables and market studies related to salaries. Based on the retrieved data, the LLM model can generate an explanation about the estimated salary, which indeed enhances the explainability of the underlying models. Validation experiments of the proposed approach conducted via calculating the R² Score confirm the accuracy and performance of the suggested salary prediction system. Ultimately, it not only solves the problem of insufficient transparency, but also increases the accuracy and speed of HR data analysis.

Fields of Study & Subjects

AI and HR Technologies Data Mining and Machine Learning Applications Explainable Artificial Intelligence (XAI)

Publication details

Journal / source
Research Digest on Engineering Management and Social Innovations
Volume
2
Issue
4
Publication date
1 April 2026
Language
EN