MODELING GROSS REGIONAL PRODUCT BASED ON CRISP AND FUZZY REGRESSIONS

Abstract

Regression modeling is a recognized tool for the analysis and forecasting of socio-economic processes. In this case, a linear regression apparatus is often used, which shows a high efficiency in solving many applied problems. In some situations the use of classical linear regression is not sufficiently substantiated. These situations are quite typical and characteristic in constructing models of socio-economic processes, especially in the case of short samples. Fuzzy regression analysis is one of the most promising areas of scientific research in this field. The fuzzy regression model gives fuzzy functional relationships between dependent and independent variables, while they can be either crisp or fuzzy. The article presents the results of a study on modeling the gross regional product as an indicator of socio-economic development of the Republic of Tatarstan by using the methods of linear crisp and fuzzy regressions. Linear GRP models constructed of four different factors (volume of shipped products, agricultural products, investment in fixed capital, and volume of performed work by type of activity “construction”). Models were built by using a sample of statistical data for 1999-2018. The forecast properties of the crisp and fuzzy regression models are compared with dividing the source sample into the training and test subsamples. The results obtained show the promise of using fuzzy regression to forecast regional economic growth.