Features of using Cox regression in various instrumental environments
https://doi.org/10.26425/1816-4277-2022-10-80-88
Abstract
The presence of large amounts of data in information and analytical systems makes it necessary to study them using machine learning and artificial intelligence methods. These models require the definition of tuning parameters related to the specifics of the subject area. The article presents a Cox regression model to solve the problem of customer churn. Cox regression is recognized as a model with high accuracy of predictions in healthcare. Therefore, it is interesting to use the model in other industries. The paper presents the results and comparative analysis of calculations on the Cox model using three tools: Statistical Package for the Social Sciences, programming language R and Russian software – analytical platform Loginom. A distinctive feature of the developed probabilistic model is the determination of the risk of event occurrence in conditions of incomplete data, as well as the identification of indicators that have a significant impact on the degree of its manifestation.
Keywords
Supporting agencies: The authors express gratitude and deep appreciation to Alexey Arustamov, General Director of CJSC Analytical Technologies, for providing software in the course of cooperation with universities, to Nikolay Paklin for recommendations and instructions in the development of components and reviewers for advice and valuable comments when working on the article.
About the Authors
I. V. Kramarenko
State University of Management
Russian Federation
Russian Federation
Inna V. Kramarenko, Cand. Sci. (Econ.), Assoc. Prof. at the Mathematical Methods in Economics and Management Department
Moscow
L. A. Konstantinova
State University of Management
Russian Federation
Russian Federation
Lyubov A. Konstantinova, Cand. Sci. (Econ.), Assoc. Prof. at the Mathematical Methods in Economics and Management Department
Moscow
References
1. Sharashova E.E., Kholmatova K.K., Gorbatova M.A., Grzhibovsky A.M. Application of survival analysis in healthcare using the SPSS statistical software package. Science and healthcare. 2017;5:5–28.
2. Shirobokova M.A., Pivkin K.S. Interpretation of models of random forest survival in application to the calculation of bank credit risk. In: Statistical methods of analysis of economics and society: Proceedings of the 11th International Scientific and Practical Conference of Students and Postgraduates, Moscow, 12–15 May, 2020. Moscow: National Research University “Higher School of Economics”; 2020. P. 163–164. https://doi.org/10.17323/978-5-7598-2305-6
3. Krasotkina O.V., Popov V.A., Nguyen T.Ch., Mottl V.V. Bayesian approach to risk factors assessment in life expectancy analysis. News of TulSU. Technical Sciences. 2013;2:168–176.
4. Izergina A.R. Review of statistical methods for assessing reliability. In: Novikov D.A. Mathematical models of modern economic processes, methods of analysis and synthesis of economic mechanisms. Actual problems and prospects of management of organizations in Russia: Proceedings of the 12th All-Russian Scientific and Practical Conference, Samara, 28–31 May 2018. Samara: Samara Scientific Center of the Russian Academy of Sciences; 2018. P. 45–50.
5. Loginom analytical platform. About the platform. https://loginom.ru/platform. (accessed 01.08.2022).
6. Sharashova E.E., Kholmatova K.K., Gorbatova M.A., Grzhibovsky A.M. Application of Cox regression in healthcare using the SPSS statistical software package. Science and healthcare. 2017;6:5–27.
7. Rey A.I., Andronova E.S. Analysis of the survival of firms in the industry: Kaplan–Meyer assessment and a model of proportional risks with time-varying variable. Certificate of state registration of the computer program No. 2020664382 Russian Federation. RU2020663158: application 28.10.2020 (Patent) 2020.
8. HR portal. Survival analysis. https://hr-portal.ru/statistica/gl14/gl14.php (01.08.2022).
9. Grishchenko D.A., Kataev A.V. Analysis of methods for modeling and forecasting customer outflow. Bulletin of Science and Education. 2018;2(5-41):21–23.
10. Paklin N.B., Katsko I.A., Kremyanskaya E.V. LTV modeling of a site subscriber using the low-code Loginom platform. SAEC. 2021;3:94–106.
11. Rumyantseva E.V., Furmanov K.K. The use of non–selective Cox-Snell residues in predicting the occurrence of events. Business informatics. 2021;1(15):7–18. https://doi.org/10.17323/2587-814x.2021.1.7.18
12. Vafeiadis T., Diamantaras K.I., Sarigiannidis G., Chatzisavvas K. Customer churn prediction in telecommunications. Simulation Modelling: Practice and Theory. Journal of Applied Technology and Innovation. 2018:1(2):7–14.
13. Yablontseva A.D. Implementation of the cox proportional risks model by means of the programming language R. Modern Science. 2021:7:380–384.
14. Loginom analytical platform. Python programming language Assistant in Loginom. https://help.loginom.ru/userguide/processors/programming/python/index.html / (accessed 20.08.2022).
15. Kaggle. Telco Customer Churn. https://www.kaggle.com/blastchar/telco-customer-churn (accessed 10.08.2022).
16. R Documentation. Cox Regression. https://www.rdocumentation.org/packages/survival/versions/3.2-10/topics/coxph/ (accessed 09.08.2022).
Review
For citations:
Kramarenko I.V., Konstantinova L.A. Features of using Cox regression in various instrumental environments. Vestnik Universiteta. 2022;(10):80-88. (In Russ.) https://doi.org/10.26425/1816-4277-2022-10-80-88
Views:
443
Email this article 