The aim of the work is to create matrix of feature objects for machine learning problems. The task is to search for data sources, develop a preprocessing algorithm, process statistical community data and form a matrix of feature objects for its further use in the clustering problem. Methods used: data analysis, finding descriptive statistics, methods of Python libraries: Pandas, NumPy. The novelty of this study lies in solving the problem of searching for relevant data of social network communities and developing an algorithm that forms a matrix of feature objects for its further use by researchers. Result: the analysis of services providing statistics of social networks was carried out, a program code was developed that implements the algorithm for generating a matrix of feature objects. The practical significance lies in the processing relevant statistical data and using a matrix of feature objects in machine learning problems.