Previsão de preço de ovos, no estado do Paraná, por meio de modelos SARIMA e PROPHET

Abstract

The present work aims to evaluate the effectiveness of time series models, in their univariate form, in forecasting the average price of 30 dozen eggs, large type, received by the farmer in the state of Paraná. The database, made available by the Institute for Applied Economic Research (IPEA), presents a historical series of prices, of 30 dozen eggs, in the period between 2006 and 2021. Forecast models based on PROPHET and SARIMA (Seasonal Autoregressive Integrated Moving Average) methodologies were implemented in Python language. Results obtained from the two models were compared. It was verified, for a period of twelve months, that the SARIMA model presented better performance, in relation to the RMSE and MAPE metrics, than the PROPHET model.