Regression methods of least deviations as a tool for enterprise analysis

Abstract

The issues of comparative analysis of models and methods from econometrics are studied using the example of fish processing enterprises. Attention is focused on the optimality criteria for regression models. Four types of two-factor problems are considered. The first two are linear with the following criteria: a) minimum sum of least squares; b) the minimum of the sum of the smallest modules. The second two are nonlinear, using the Cobb-Douglas function as an example: a) with multiplicative deviations and the criterion of the minimum sum of squared logarithms of relative deviations; b) with additive deviations, the problem is internally nonlinear, with the minimum sum of least squares criterion. Analytical and search methods of solutions are used.