Spatial unevenness of costs formation in the implementation of transport and pedestian correspondences in large city

The article considers the issues of spatial unevenness of the structure of mobility of city residents. The purpose of the study is to increase the accuracy of forecasting the parameters of the transport system based on data analysis and differentiation of corre-spondence of city residents using machine learning methods. The study puts forward a hypothesis that the structure of transport demand is distributed unevenly across the city and depends on the costs of implementing correspondence. To study the corre-spondence of city residents, the article uses a transport macromodel of the city of Tyumen. An analysis of the conditions for the formation of costs for the implementa-tion of correspondence, which are recorded in the macromodel, is provided. To ana-lyze the correspondence and assess the impact of the parameters of the city's transport system using machine learning, clustering of correspondence was carried out, which are grouped into 5 clusters. The results of data processing in the transport model showed that the distance and travel time have the greatest impact on the share of travel on individual and public transport. For two clusters, with an increase in distance and travel time, the share of trips on public transport decreases, for the other two, on the contrary, it increases. In this case, the general form of the dependence for all corre-spondences is described by the parabola equation. The formation of 5 correspondence clusters is due to the different levels of transport services, which are determined by the level of development of the street and road network, public transport and its infra-structure, taking into account the actual division of the city into 4 parts by the river, the Trans-Siberian Railway and Fedyuninsky Street. The further direction of research is related to determining the most significant factors for each cluster and increasing the arbitrariness of working with transport macromodels due to the development of digital twins of urban transport.