Análise comparativa de modelos de previsão para receitas correntes mensais do estado de Alagoas: abordagens econométricas e redes neurais

Abstract

In this research, a comparative analysis was conducted between econometric and neural network models to predict current revenues in Alagoas. Using data from the Transparency Portal, the SARIMA models and the NNAR Neural Network were applied. The results highlighted the superiority of NNAR in forecasting current revenues and ICMS, while SARIMA demonstrated better performance in forecasting current transfers. These results have important implications for financial planning in Alagoas, emphasizing the usefulness of these models in accurately forecasting public revenues. However, it is essential to recognize the limitations inherent in the available data. Future investigations are recommended that explore additional techniques and variables to improve forecast accuracy, contributing to more informed financial management in the state context.