Previsão de mercado: competição de métodos aplicada à previsão do consumo de energia elétrica nas regiões do Brasil

Abstract

This work seeks to understand the behavior of the regional electricity market and has as its main objective study electricity consumption by regions in Brazil and forecast this market in the short term. For this purpose, statistical analysis applied to time series was used. Two forecasting methods were applied, the Exponential Smoothing Method and the Box & Jenkins Method, with a competition between them to select the winning model. The Box & Jenkins Method was defined as the winner, as it had better predictive performance. The estimated models projected growth in electricity consumption in the short term in each region.