On the use of neural networks to solve the social clustering problem

The problem of social clustering being studied in the paper is one of the main subtasks; its solution is an integral part of analysis and prognosis of socio-economic processes. Analysis and systematization of knowledge in the field of applying neural network modelling to regional system social clustering problem solving are implemented. It was demonstrated that today, the main factor of economic growth is human capital, which is composed of quantitative and qualitative features. The main quantitative element is population replacement which has a bearing on human capital development sustainability. Qualitative component has several aspects in it: healthcare, culture, education and science are among them. To estimate human capital structure, the population is divided into social clusters by these aspects. It was also shown that since social cluster is an attribute of sociogenesis, processes of social clustering themselves are the result of people social interactions. Social cluster is a specific state of social entity which includes description of not only entity's objects, but the processes which led to its structural development and interactions with social environment. As part of the study, a conclusion was made that neural networks enable one to apply cluster analysis to the society. Neural networks prove notable capabilities to solve poorly formalized tasks; they are resistant to frequent environmental changes and effective to use when working with a large amount of incomplete or contradictory information. While studying the issue, it was observed that structural and statistical features of social clusters reflect aggregation of their elements. The structure of a social cluster is a characteristic which represents a conjunction of stable connections which provide its unity. Under different external and internal changes, the main properties of social clusters are preserved. The grading of social demographic elements by health condition and cultural and educational level is set, in accordance with which collecting a statistical data to solve the clustering problem is implemented.