IQTISODIY O‘SISHNING IJTIMOIY SAMARADORLIGINI NEYRO-RAVSHAN EKONOMETRIK MODELLASHTIRISH

The article presents a multifaceted econometric approach to modeling the social effectiveness of economic growth-the classical econometric model, the hybrid model, and the neuro-explicit model and provides a comparative analysis of their results. The classical econometric model is constructed in the form of multi-factor linear regression (OLS), in which, along with the growth rates of investment and employment, the growth rate of gross domestic (regional) product (GDP) is taken as the input parameter. The empirical basis of the study is quarterly indicators (64 observations) covering the period from 2010 to 2025 using the Kashkadarya region as an example. The hybrid model is built by correcting the classical regression prediction at the residual level using a neural network, while the neuro-clear model is built by training a three-input (3–5–5–1 architecture) neural network and extracting clear rules from it using a decision tree. According to the research results, all three models possess high accuracy (RMSE 0.729; 0.693; 0.575), where the neuro-clear model provides the highest predictive accuracy, while the classical model provides the highest explainability. Based on the classical econometric model, a statistically significant and positive impact of GDP growth rates on the social efficiency of the Kashkadarya population was determined (β = +0.43; p < 0.001)

Hali tarjima qilinmagan