THE IMPACT OF ECONOMIC INDICATORS ON CONSUMPTION: A NEURAL NETWORK APPROACH

Economic indicators are fundamental for understanding consumption patterns and sectoral performance, forming the basis for informed decisions in public policy, investment strategies, and business planning. This research employs neural networks to model the complex and nonlinear relationships between economic variables and consumer behavior. A neural network architecture with two dense layers of 64 neurons each and a regression output layer was implemented using TensorFlow and Keras. The models achieved high predictive accuracy, with R² values exceeding 0.90 for most indicators, demonstrating their ability to explain substantial variability in the data. Variations in Mean Absolute Error (MAE) across indicators reflected specific economic characteristics, further confirming the robustness of the methodology. The findings underscore the potential of neural networks to overcome the limitations of traditional econometric models by capturing complex, multivariate interactions. This approach represents a significant advance in forecasting economic variables, enabling the identification of patterns and trends that inform sector-specific strategies. Future research could enhance predictive capabilities by incorporating broader datasets, domain-specific variables, and advanced architectures, fostering hybrid frameworks that balance accuracy with model interpretability for both academic and practical applications in economic analysis.