Реализация конкурса регрессионных моделей с применением критерия согласованности поведения

Сергей Иванович Носков

doi:10.17308/sait.2021.2/3511

Authors

Sergey I. Noskov Irkutsk State Transport University https://orcid.org/0000-0003-4097-2720 (unauthenticated)

DOI:

https://doi.org/10.17308/sait.2021.2/3511

Keywords:

regression model, adequacy criteria, least squares methods, moduli, anti-robust and mixed estimation

Abstract

The paper describes the main stages of the implementation of the competition of regression models of complex objects. It consists in constructing a set of alternative options for a model of a given class and then choosing the best option based on the use of a vector criterion for assessing adequacy. It includes, in particular, the Fisher, Student, Durbin-Watson criteria, multiple determination, bias, informativeness of a set of independent variables, residual variance, average relative errors of approximation and forecast, width of the domain of the model. In this case, the total number of potential variants of the model can amount to tens of thousands for dimensions corresponding to real situations that arise during modeling. The main stages of organizing a competition of models are described: construction of a set of admissible options according to their compliance with the meaningful meaning of factors and the values of particular criteria of adequacy; maximization on this set of the generalized criterion for the consistency of the behavior of the calculated and actual values of the dependent variable; in the case of non-uniqueness of the solution to this problem, the implementation of the ideal point method on the generated set of variants of the model. The model obtained as a result of the competition will correspond to the meaning of all its constituent variables and be of high quality.

Author Biography

Sergey I. Noskov, Irkutsk State Transport University

Doctor of Technical Sciences, Professor, Professor of the Department of Information Systems and Information Security, Irkutsk State University of Railways

References

Noskov S. I. (1996) Object modeling technology with unstable operation and data uncertainty, Federal State Public Educational Establishment of Higher Training Eastern Siberia Institute of the Ministry of the Interior of the Russian Federation, 110, Lermontov Street, Irkutsk. 320 p.

Noskov S., Potorochenko N. (1992) System for the implementation of the “competition” of regression dependencies // Control systems and machines. No 2/4. P. 111–116.

Bazilevsky M. P., Noskov S. I. (2009) Technology of organizing a competition of regression models // Information technologies and problems of mathematical modeling of complex systems. No 7. P. 77–84.

Bazilevsky M. P. (2018) Nonlinearity Criteria for Quasilinear Regression Models // Modeling, Optimization and Information Technologies. V. 6. No 4 (23). Р. 185–195.

Bazilevsky M. P. (2020) Fundamental block of algorithms for constructing well-interpreted qualitative regression models // Information technologies and mathematical modeling in the management of complex systems. No 3 (8). Р. 1–10.

Bazilevsky M. P. (2017) Development and research of algorithms for estimating the parameters of additive power regression // Modern technologies. System analysis. Modeling. No 4 (56). Р. 131–138.

Bazilevsky M. P. (2012) Software and mathematical support for automation of multicriteria choice of regression models: abstract of ... dis. can. those. sciences. Irkutsk. 27 p.

Bazilevsky M. P. (2016) Algorithm for choosing the best regression in terms of the coefficient of determination with an adjustable effect of multi-collinearity between regressors // Information technologies and problems of mathematical modeling of complex systems/ No 15. P. 31–38.

Eliseeva I. I., Kurysheva S. V., Kosteeva T. V. and others (2007) Econometrics. Moscow: Finance and Statistics. 576 p.

Fox W. P., Bauldry W. C. (2020) Advanced Problem Solving Using Maple. LocationBoca Raton. 404 p. DOI https://doi.org/10.1201/9780429469626.

Kissell R. L. (2021) Algorithmic Trading Methods (Second Edition) // Applications Using Advanced Statistics, Optimization, and Machine Learning Techniques. P. 151–173.

Pardoe I. (2021) Applied Regression Modeling. John Wiley & Sons, Inc. 310 p.

Guerard J. B., Gultekin A. S. RegressionAnalysis and Estimating Regression Models // Quantitative Corporate Finance. P. 263–300.

Westfall P. H., Arias A. L. (2020) Understanding Regression Analysis. A Conditional Distribution Approach. LocationBoca Raton. 514 p.

Draper N, Smith G. (2016) Applied Regression Analysis, 3rd ed. Williams. 912 p.

Noskov S. I. (2020) Method of antirobast estimation of linear regression parameters: number of maximum on the module of approximation errors // South – Siberian scientific bulletin. No 1(29). Р. 51–54.

Noskov S. I. (2019) About the method of mixed estimation of parameters of linear regression // Information technology and mathematical modeling in the management of complex systems. N. 1. P. 14–20.

Lotov A. V., Bushenkov V. A., Kamenev G. K., Chernykh O. L. (1997) The computer and the search for a compromise. Method of achievable goals. Moscow : Nauka. 239 p.

Noskov S. I. (1994) Construction of econometric dependencies taking into account the criterion of “consistency of behavior” // Cybernetics and Systems Analysis. No 1. P. 177.

Noskov S. I. (2018) Generalized criterion of coordination of behavior inregression analysis // Information technology and mathematical modeling in the management of complex systems. No 1. P. 14–20.