AI-BASED NOWCASTING OF REAL INCOME AND PROFIT GROWTH IN UZBEKISTAN USING MULTI-SOURCE ECONOMIC DATA

Abstract

The study explores a data-driven approach to anticipating short-term changes in Uzbekistan’s real income growth by integrating national and international economic indicators. Using open datasets from the Statistics Agency, the World Bank, and nowcasting sources, the research develops a predictive framework that applies machine learning algorithms—specifically Random Forest and LSTM models—to capture the dynamic patterns of income fluctuations between 2010 and 2024. Unlike conventional econometric techniques that rely on lagged data, the proposed method allows near–real-time estimation of income growth trends. The results suggest that foreign direct investment, technological exports, and trade openness exert the strongest influence on income dynamics, while energy intensity demonstrates a moderate inverse relationship. The model’s empirical performance indicates that combining AI-based forecasting with official statistics can improve the accuracy and timeliness of socio-economic monitoring in transition economies such as Uzbekistan.