UDC 512.2
MODELS OF TEMPORAL REGRESSION DEPENDENCIES AND THEIR APPLICATION TO LIVING STANDARD OF POPULATION
V. K. Klochko, Dr.Sc. (Tech.), full professor, Department of automation and information technologies in control, professor of the department, RSREU, Ryazan, Russia;
orcid.org/0000-0003-2550-999X, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
V. P. Kuznetsov, Ph.D. (Tech.), associate professor, Department of automation and information technologies in control, associate professor of the department, Ryazan, Russia;
orcid.org/0000-0000-0000-000X, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
In traditional statistical analysis, the construction of regression models assumes their stationary nature. In this case, the selection of data for building models does not reflect the time of data reception. The aim of the work is to expand the possibilities of using regression models in forecasting by including time in them. In addition, obtaining a sample of measurements at different points in time increases its total volume, which allows solving the problem of small samples when building regression models. The paper proposes two approaches to constructing time regression models. In the first approach, time is included in the system of regression equations along with the data obtained at different points in time. In the second approach, the regression coefficients found independently at different points in time are subject to time processing. As an application, linear regression models of the dependence of poverty level on average population income of the regions of Russian Federation and on the year number were obtained. The model data show a more uniform change in regression dependencies over years compared to a traditional model. The results can be used in Rosstat calculations.
Key words: regression equations, temporal dependencies, forecasting, poverty level, incomes of the population.