Improving accuracy of approximation of economic regularities by polynomials of high degrees by limiting amplitude of variations

Approaches to a formalized description of patterns of change in economic indicators in relation to dynamic series and correlation fields, including multidimensional ones, are given. Models are built on the basis of high-degree polynomials with artificial limitation of the amplitude of oscillation. Theoretical prerequisites for evaluating the sensitivity of the resulting functions to the number of control levels are presented, technologies for introducing restrictions are developed, an analogue of the Cheddock scale is proposed as a ratio of the lengths of the semi-axes of the scattering ellipse.