Selection of сategorical parameters in modeling systems

Abstract

Annotation. The goal is to develop a methodology for working with factors that have several categorical levels to select one of the levels as a parameter of the simulated system. The technique . To select a parameter, statistical data are used, obtained either as a result of observations or as a result of a reusable simulation model run and, therefore, methods of mathematical statistics are used to achieve the goal. A multiple regression model is considered for categorical factors, the levels of which are presented as dummy variables. Results. The application of the proposed method allows not only to assess the effect of the factors and carry out a pair-wise comparative analysis of their levels, but also to determine one level of each categorical factor that is the best under these conditions. Scientific novelty . The proposed method makes it possible to reduce the problem of choosing a category parameter of the system being modeled to a regression analysis problem with subsequent testing for optimality of the regression function. The final choice of the parameter as one of the category levels in the case of two factors is found as a solution to the problem of nonlinear programming. Practical significance . The choice of parameters in the preparation of the system model is one of the main stages. There is no particular problem when it comes to parameters that take numerical values: for this purpose, the methods of mathematical statistics for testing hypotheses of expectation and two-sample criteria are used. In the case when the factor has several categorical (non-numeric) levels, dispersive analysis is used to analyze their influence, which makes it impossible to solve the problem with parameter selection. The proposed method allows you to make such a choice.