Identification of econometric models of simultaneous equations and control systems based on machine learning

<jats:p>The chapter proposes a method for analyzing economic systems based on the use of a model of simultaneous equations and machine learning. The architecture of a fully connected neural network capable of reflecting the structure of a discrete-time control system and allowing the identification of a system of simultaneous equations is considered. A method for training a neural network is proposed that allows estimating the coefficients of a set of simultaneous equations. A method for forecasting a multivariate time series based on the estimated structural parameters of the system using a perceptron is presented.</jats:p>