Agent modeling in research and forecasting of socio-economic systems and processes

All-pervading digitalization, which is a process of digitizing not only quantitative but also qualitative aspects of objects in the real world, creates new challenges and new opportunities for humanity. The embodiment of such opportunities, in particular, is the use of the growing volume of data in science. And here we are talking not only about classical statistical methods, but also about complex neural network and simulation approaches, many of which are successfully used today. This work is devoted to the study of the experience of using one of such approaches agent-based modeling. This methodology is universal and allows you to reproduce socio-economic systems and processes of varying degrees of complexity. Its application may be limited only by the theoretical knowledge of the researcher and / or the computing power of the platform for the implementation of simulation models. The introductory part of the work discusses the conceptual foundations of the agent-based approach methodology and certain aspects of its advantages in the study of social processes. An approach to the classification of agent-based models according to the aggregation-atomization and abstraction-concretization scales is presented. The second part is devoted to the analysis of the experience of using agent-based modeling in the field of socio-economic relations. The review includes the works of foreign and Russian authors. In conclusion, the trends in the development of agent-based modeling and its prospects are discussed. Based on the analysis of the dynamics of the number and the thematic structure of studies carried out using the methodology of agent-based modeling, it is concluded that the latter is expanding into the subject areas of the humanities.