Resumen
The enhancement of Big Data, computer systems for data aggregation, distributed information systems determines the need to create new decision-making systems. Decision support systems at the federal level should not only increase the quantitative characteristics of data processing but also receive new types of generalized models and analytical sources. Along with the potential of large data volumes processing and machine learning, approximate models of dynamic systems are becoming widespread for financial time series. The development of approximate models makes it possible to qualitatively assess the dynamics of changes in parameters, to build forecasts in the form of intervals that are insensitive to local fluctuations in financial instruments. The paper describes the formation of a decision support system based on the development of estimates of the approximate systems dynamics in the conditions of Big Data when targeting the balance of funds on the single treasury account of The Federal Treasury of The Ministry of Finance of the Russian Federation. The main conceptual architectural solutions for the development of a decision support system are given.