Interval estimates of the solution statistical model of management

L. V. Tsibriy, S. S. Muliar

Abstract


Goal. Create a model of managing a complex system based on statistical data and find the optimal management solution. Method. The construction of a statistical model of a complex system that reflects the whole range of its parameters and relationships is the subject of many studies. All of them are based on methods of mathematical statistics and lead to the model of multiple regression [1,2,4]. The aim of the simulation is not only to evaluate the parameters of the system's operation, but also to manage it. This leads to the need to solve the problem of stochastic programming [3]. There is no single analytical method for solving such problems.
The paper outlines one of the approaches that allows solving the problem of finding the best control with the help of the simulation model considering the consideration of interval regression estimates as a random effect provided by the mathematical model of stochastic programming. This allows us to find not only point estimates of the deterministic optimal solution, which occurs when reducing the problem of stochastic programming to the problem of nonlinear programming. Results. The application of the proposed method leads to the receipt of interval estimates of the controlled variables and the target function of the optimization problem, corresponding to the interval estimates of the regression of the explained parameters on the explanatory parameters. Scientific novelty. A method for solving the control problem of a complex system is proposed, which makes it possible to take into account the stochastic nature of the model and to find not only point, but also interval estimates of the optimal solution. Practical significance. Interval estimates of the decision of the statistical model, taking into account the random spread of statistical data, are necessary for making the right decision when choosing the parameters of the system being designed, which will allow to take into account possible undesirable random effects.

Keywords


statistical model; multiple regression; stochastic programming; simulation model; interval estimates; optimal management1. Tsibrij L.V. Vvedenie v statisticheskij analiz [Introduction to statistical analysis]. Dnepropetrovsk: PGASA, 2016, 188 p. (in Russian). 2. Muliar S.S, Tsybrii L.V. Rozrobka alg

References


Tsibrij L.V. Vvedenie v statisticheskij analiz [Introduction to statistical analysis]. Dnepropetrovsk: PGASA, 2016, 188 p. (in Russian).

Muliar S.S, Tsybrii L.V. Rozrobka alghorytmu pobudovy statystychnoi modeli optymalnoho upravlinnia [Development of optimal-management statistical model]. Stroitel'stvo, materialovedenie, mashinostroenie [Construction, Materials Science, Mechanical Engineering]. Komp'yuternye sistemy i informacionnye texnologii v obrazovanii, nauke i upravlenii [Computer systems and information technologies in education, science and management]. Dnepropetrovsk: PGASA, 2016, iss. 94, pp. 119-124. (in Ukrainian).

Ermol’ev Yu.M. Metody stoxasticheskogo programirovanya [Methods of stochastic programming]. Optimizaciya i issledovanie operacij [Optimization and operations research]. Moskva: Nauka, 1976, 240 p. (in Russian).

Levin D.M., Stefan D., Krebill T.S. and Berenson M.L. Statistika dlya menedzherov s ispolzovaniem Microsoft Excel [Statistics for managers using Microsoft Excel]. Ed. 4, Moskva: Vil’yams, 2004, 1312 p. (in Russian).

Tomashevskyi V.M. Modeliuvannia system [Systems’ modeling]. Кyiv: BHV, 2005, 352 p. (in Russian).

Dubrov A.M. Mkhitaryan V.S. and Troshin L.I. Mnogomernye statisticheskie metody [Multidimensional statistical methods]. Moskva: Finansy i statistika, 2000, 350 p. (in Russian).


GOST Style Citations