About the article
N.Yu. Mutovkina, A.N. Borodulin
An adaptive system of neuro-fuzzy inference is considered, which allows determining a fair price for a product with improved consumer properties in conditions of monopolistic competition. The input variables of the system are indicated (the key consumer characteristics of the products on which the sales price depends – the output variable). As an initial statistical sample, data on the properties of such products and the prices set for them by manufacturers were taken. As a tool for creating an appropriate model of such a neuro-fuzzy system, the ANFIS subsystem of the MATLAB modeling environment is considered and used. In the process of modeling, the main characteristics affecting the selling price of the goods are determined, a formalized statement of the problem is given. The architecture of the corresponding fuzzy inference system in the form of the ANFIS neuro-fuzzy network, which implements a Sugeno-type fuzzy inference system, has been developed and described. The sequence of development and computer implementation of the model is presented in detail. An example of practical calculations based on the developed model is given, the main results are analyzed.
pricing, hybrid neural networks, expert assessments, membership functions, ANFIS editor, fuzzy inference.