Predicting the viability of non-public railway track connection projects using ensemble machine learning methods
https://doi.org/10.46973/0201-727X_2026_1_59
Abstract
The paper addresses the scientific and practical problem of increasing the objectivity of selecting projects for connecting non-public railway tracks to public infrastructure. A method for predicting the viability of project implementation based on ensemble machine learning methods is proposed. The scientific novelty lies in the development and experimental validation of predictive models using multi-criteria project indicators as feature space, as well as in the application of SHAP analysis for result interpretation. The practical significance of the work is the creation of algorithmic support that reduces the share of inefficient investments through preliminary assessment of risks of failing to achieve declared traffic volumes.
About the Author
M. O. Fonseca
Russian University of Transport (RUT MIIT); orporate Transport Service Center – JSC "Russian Railways"
Russian Federation
Russian Federation
Fonseca Marina Olegovna, Chair "Digital Technologies for Transport Process Management", Postgraduate Student; Head of the Department for Automation and Implementation of Information Systems
References
1. Fonseca, M. O. Problems of interaction with owners of non-public railway tracks after commissioning of newly built or reconstructed tracks / M. O. Fonseca, V. E. Nutovich // Digital transformation of transport : challenges and prospects : proceedings of the IV International scientific and practical conference (Moscow, September 25, 2024). – Moscow : Russian University of Transport (MIIT), 2024. – P. 127–130.
2. VCIOM. Assessment of the quality of freight railway transport services : report on the results of a user survey. – Moscow, November 2023. – 89 p.
3. On approval of the Rules for consideration of applications of legal entities or individual entrepreneurs owning on the right of ownership or other right of non-public railway tracks, on the connection of non-public railway tracks under construction, reconstruction, or restoration to реконструируемых, восстановленных железнодорожных путей необщего пользования к инфраструктуре ОАО «РЖД» или о развитии существующих примыканий : распоряжение ОАО «РЖД» от 03.10.2022 № 2542/р. – Москва, 2022. – 88 с. the infrastructure of JSC «Russian Railways» or on the development of existing connections : Order of JSC «Russian Railways» No. 2542/rdated 03.10.2022. – Moscow, 2022. – 88 p.
4. On introducing amendments to Order of JSC «Russian Railways» No. 2542/r dated October 3, 2022 : Order of JSC «Russian Railways» No. 1929/r dated 01.08.2023. – Moscow, 2023.
5. On approval of the Rules for non-discriminatory access of carriers to the infrastructure of public railway transport, consignors to the service of freight transportation by public railway transport : Decree of the Government of the Russian Federation No. 710 dated 25.11.2003 (as amended on 27.05.2024). – Moscow, 2003. – Access from the legal reference system «ConsultantPlus».
6. Larchenko, D. V. Intelligent decision support systems in transport process management / D. V. Larchenko, L. A. Fedorov. – Moscow : INFRA-M, 2019. – 198 p.
7. Fonseca, M. O. Automation of coordination of conditions for connection of new non-public railway tracks to public railway tracks / M. O. Fonseca, V. E. Nutovich // Digital transformation of transport : challenges and prospects : proceedings of the III International scientific and practical conference (Moscow, 2023). – Moscow : Russian University of Transport (MIIT), 2023. – P. 10–13.
8. Russell, S. Artificial Intelligence : a modern approach / S. Russell, P. Norvig ; trans. from English. – 4th ed. – Moscow : Williams, 2021. – 1416 p. (Chapter 19: “Ensemble-Based Learning”).
9. Molnar, K. Interpretable machine learning : a guide for making black box explainable models / K. Molnar ; trans. from English. – Moscow : DMK Press, 2022. – 382 p. (Chapter on SHAP).
10. GOST R 59276-2020. Artificial intelligence systems. Methods for ensuring trust. General provisions. – Moscow : Standartinform, 2021. – 16 p.
11. Description of the functional capabilities of the Automated System of the new generation unified technological process. Technical conditions module (AS ETP NP.TU) : internal regulatory document of JSC “Russian Railways”. – Moscow, 2023.
12. System for integrated modeling of station and section operation / A. M. Zamyshlyaev, S. V. Kalinin, D. Yu. Khalevin [et al.] // Proceedings of JSC «Research and Design Institute for Information Technology, Automation and Communication in Railway Transport». – 2021. – Issue 11, Vol. 1. – P. 90–98.
Review
For citations:
Fonseca M.O. Predicting the viability of non-public railway track connection projects using ensemble machine learning methods. Vestnik Rostovskogo Gosudarstvennogo Universiteta Putej Soobshcheniya. 2026;(1):59-65. (In Russ.) https://doi.org/10.46973/0201-727X_2026_1_59
Views:
33
JATS XML
Email this article 