Comparative study between ARIMA and ETS forecast models for temporal data on milk production in Brazil

Abstract

Milk production is constantly growing, as it moves the economy and is a source of income for several families. Effective planning of activities performed by both milk and dairy producers is directly related to expectations regarding the annual production of milk. The estimation of milk production can be performed using numerical-statistical forecasting models, with the help of software such as R. Thus, this article addresses a comparative analysis of the forecast of industrialized milk production in Brazil, using ARIMA (Autoregressive Integrated Moving Average) and ETS (Error, Trend, Seasonal) models. The determination of model’s parameters and other statistical calculations were performed using free software R for monthly and quarterly data series on milk production, obtained from the IBGE website, from 2004 to 2018. The models provided forecasts for the year 2019 and these were compared with actual values. The metrics used were MAPE (Mean Absolute Percentage Error), RMSE (Root Mean Square Error), and MAE (Mean Absolute Error), which indicate that the ARIMA model presented greater accuracy for both analyzed series.