Evolving Fuzzy Modeling with Variable Step-Size applied to Gas Prices Forecasting in Brazil
Eduardo Ravaglia Campos Queiroz, Kaike Sa Teles Rocha Alves, Fernando Luiz Cyrino Oliveira, Eduardo Pestana de Aguiar · 2021
Fuel represents one of the main transport costs and, consequently, of a logistical operation. So, a computational tool that allows reliable forecasts on the fuel prices becomes a competitive differential for the logistics operator, especially in a country of continental dimensions such as Brazil. The present study developed a participatory evolving fuzzy model, with a variable step-size, denominated VS-ePL-KRLS, applied to the forecast of weekly prices of diesel oil S500 and S10, at the national level, for biweekly and monthly horizons. The proposed model presented better accuracy in relation to the existing ones of this class in the literature, without loss of computational performance for all four analyzed scenarios.