CONSTRUCTION OF A HOMOGENEOUS NESTED PIECEWISE LINEAR REGRESSION WITH LAGGING VARIABLES

When constructing regression models of objects of any nature, it is often necessary to use nonlinear approximating constructions, including piecewise linear ones, while the process under study can have a pronounced dynamic nature, therefore, lagging (lag) variables can be used as regressors. The research purpose is to develop an algorithmic method of identifying the parameters of a homogeneous nested piecewise linear regression with lagging variables. Materials and methods. To achieve the goal, the methods of reducing the problems of estimating the parameters of nested piecewise linear models to linear Boolean programming problems proposed earlier by one of the authors were used. The least absolute values ​​method, known in regression analysis, was also applied. The volume of oil production in the Russian Federation was adopted as the modeling object using statistical initial data in 2013–2022. Research results. An algorithmic method of constructing a homogeneous nested piecewise linear regression with lagging variables has been developed, which is reduced to solving a linear Boolean programming problem. It has been applied to construct a model of the possible volume of oil production in the Russian Federation. In this case, data on the volume of capital investments of Russian vertically integrated oil companies and on the commissioning of new wells have been used as independent variables. Conclusions. The developed method of constructing a homogeneous nested piecewise linear regression with lagging variables using the least absolute values ​​method is reduced to a linear Boolean programming problem. Such models allow us to identify limiting values ​​of dependent variables, taking into account possible delays in influence at external and internal levels.