METHOD OF STATISTICAL LEARNING OF THE PARAMETERS OF THE EXPONENTIAL MATHEMATICAL MODEL

Abstract

In the practice of data processing in many applied fields: control of production processes; automation of dynamic object management; applied economic analysis - strict requirements are put forward for the accuracy, efficiency and reliability of the assessment and forecasting of the development of the researched processes. This requires the complication of mathematical models used in statistical learning technologies in the direction of increasing their nonlinearity. This is possible in the development of methodological support for Data Science technologies with the introduction of analytical approaches to determining the parameters of nonlinear models.

Analytical method of determining parameters for nonlinear models using statistical samples in Data Science is suggested here. It is based on statistical learning methodology in scheme of differential non-taylor transformations. As an example of suggested method usage, the exponential model was built. The effectiveness of its application was confirmed using classical numerical iterative calculations as comparison.