Gross domestic product is a macroeconomic indicator reflecting the market value of all final goods and services produced during the year in all sectors of the economy in the territory of a given state. It is very significant for the economy as a whole and makes it possible to assess the dynamics of economic growth in the country. The main goal of developing a computerized system for forecasting gross domestic product (GDP) is the development of scientific and theoretical and methodological bases for calculating and forecasting dynamics of real GDP based on macro-economic models. This article describes a causal model for predicting GDP using linear regression and the apparatus of neural networks, as well as a comparative analysis of the methods presented.