For example,Бобцов

Neural network modeling of the relationship between the investment potential of Russian regions and their ESG ranking

Annotation

The purpose of this article is to identify, test and evaluate the impact of ESG ranking on the investment potential of Russian regions (subjects of the Russian Federation) based on neural network modeling. The sample used was data from 83 Russian regions for 2020. A cross-regional multiple regression analysis was conducted using multilayer perceptron neural network models. To identify causal relationships between the investment potential variable and three regressors: regional ESG ranking, regional investment risk and Gross Regional Product, the regression application of the STATISTICA (StatSoft) analytical software package was used. The quality and accuracy of the multilayer perceptron regression model was assessed by calculating the Mean Absolute Percentage Error. The numerical experiments showed that the neural network modeling method is available in implementation, does not require large computing power and demonstrates good accuracy. In the course of neural network modeling of nonlinear regression of investment potential of the subjects of the Russian Federation, a reliable MLP 3-4-1 model was specified with high accuracy describing and predicting the target variable - the investment potential of the region. The results obtained confirm the hypothesis put forward in the article: the investment potential of Russian regions depends to a greater extent on the ESG ranking and to a lesser extent on the investment risk of the region and the Gross Regional Product. Global sensitivity analysis showed that all constructed neural networks define the ESG variable as the most important. The proven priority of the importance of the influence of the ESG ranking on the investment potential of the region can be used by regional authorities to justify priority areas of regional policy to increase the investment attractiveness of the subjects of the Russian Federation.

Keywords

Постоянный URL

Articles in current issue

CONTENT © ITMO UNIVERSITY 1993-2025