IMPROVING THE QUALITY OF TRANSPORTATION BASED ON THE CONSTRUCTION OF COMPLEX PREDICTIVE MODELS OF INFRASTRUCTURE LOADING

Abstract

Accurate construction of time series forecasts is a key element in the system of support and management decision-making. This article presents a method of multi-stage system forecasting of time series. The effectiveness of the proposed method is experimentally justified by the example of the arrival of car traffic at the junction points of railways. The most important contribution is the introduction of a system forecast, in which the methods of forecasting traffic flows will be mutually consistent and complementary, since the use of statistical methods alone will not fully reflect all the changes that occur in the transport complex of the Russian Federation. This hybrid combination provides more competitive forecasts compared to other methods. Moreover, such a hybrid model is easier to interpret by decision makers when modeling trend series.