This study examines a pressing issue: companies' need for predictive analytics technologies to manage employee turnover. Manual analysis of HR reports is time-consuming, relies on the experience of HR specialists, and is susceptible to human error. Intelligent systems based on deep learning models can predict individual and group layoff risks, identify key factors, and generate retention recommendations, improving planning accuracy and response times. This study focuses on employee turnover management at Russian Railways. It also examines machine learning methods and algorithms used for predictive analysis of tabular HR data. The objective of the study is to develop an intelligent subsystem for predicting employee turnover risks and generating recommendations for mitigation, designed to assist HR specialists and branch managers. To achieve this goal, we analyzed the operations of the research facility, the composition and structure of the existing information system, and the prospects for using artificial intelligence technologies. The designed subsystem is intended to support HR specialists and managers and does not replace their decisions: final responsibility remains entirely with the specialist. The subsystem is designed to load data from an existing information system, calculate individual and group layoff risks using the TabNet model, identify key risk factors, generate targeted recommendations, and generate structured reports. The results are integrated into EK-ASUTR and corporate dashboards for subsequent analysis. TabNet, a deep neural network specifically designed for tabular data, is used as the intelligent solution. TabNet uses a sequential attention mechanism, which dynamically selects the most informative features at each decision step. This ensures sparse use of features and increases the generalization ability of the model. The neural network was trained using a historical dataset of HR data from the EK-ASUTR of JSC Russian Railways for the period 2015-2024, designed to solve the problem of binary classification of the risk of employee dismissal within the next 12 months. An assessment of the subsystem's implementation effectiveness was conducted, confirming the optimization of HR processes, a reduction in analysis time (by 70-80%), and an increase in forecast accuracy.
MATHEMATICAL MODEL, NEURAL NETWORK, MACHINE LEARNING, HR SPECIALIST
1. Federal Law "On Personal Data" dated July 27, 2006 No. 152-FZ // Federal Law "On Personal Data" dated July 27, 2006 No. 152-FZ (latest revision) \ ConsultantPlus
2. Standard of JSC Russian Railways 1.18.002-2009 Information Security Management// STO RZD 1.18.002-2009 Information Security Management. General Provisions / 1 18 002 2009
3. I.N. Kalinovskaya Management in Social and Economic Systems, 31, 28-30 (2022)
4. A.F. Denisov, D.S. Kardash, Economics and Management, 6 (152), 26-37 (2018)
5. M. G. Nuriev, E. S. Belashova, K. A. Barabash Software systems and computational methods, 1, 1-12. (2023) – DOIhttps://doi.org/10.7256/2454-0714.2023.1.39547. – EDN SNAYLQ.
6. Yu.N. Smirnov, A.I. Abdullin, A.V. Kalyashina, Scientific and Technical Herald of the Volga Region. 6, 106-109.(2023) – EDN JAZFDJ
7. A. R. M. Khudkhair, E. V. Zargaryan, Yu. A. Zargaryan Bulletin of SFedU. Technical sciences, 227, 3, 211-222 (2022) - DOIhttps://doi.org/10.18522/2311-3103-2022-3-211-222.
8. M.A. Rukomin, Bulletin of Science, 89, 8, 348-354. (2025)
9. V.I. Simashev, M.G. Nureyev, International Research Journal, 154, 4, 123-124 (2025) - DOI:https://doi.org/10.60797/IRJ.2025.154.89.
10. N. V. Vlasova, V. A. Olentsevich, Modern Technologies. Systems Analysis. Modeling, 77, 1, 127-135. (2023) – DOIhttps://doi.org/10.26731/1813-9108.2023.1(77).127-135
11. M. G. Nureyev, M. G. Lapteva, V. A. Kosmynin International Research Journal, 158, 8. (2025) – DOIhttps://doi.org/10.60797/IRJ.2025.158.44.
12. A. A. Vasiliev, E. V. Furego, A. A. Lyubchenko, S. V. Eroshenko, The Road and the Road Economy, 10, 2-5 (2024)
13. A.V. Kalyashina., Herald of Technological University. I, 27, 12, 147-152 (2024)https://doi.org/10.55421/1998-7072_2024_27_12_147
14. A.V. Kalyashina, Yu.N. Smirnov Scientific and Technical Bulletin of the Volga Region, 11, 119-122 (2023) – EDN LXULQC
15. A. Khamitov, E. Yu. Belova, Information Technologies. Problems and Solutions, 25, 4, 27–33. (2023)
