Regressão múltipla e o modelo ARIMA na previsão do preço da arroba do boi gordo

Abstract

The livestock has an important function in the worldwide economy. It is ranked as one of the main activities responsible for the animal protein production, which is consumed especially through beef and milk. This provides substance to various sections of the economy, for instance: organic manure, animal food products, products for the footwear industry as well as for the clothing industry, pharmaceutical products and many others. Further on, in many cases the livestock is animal’s work used to agricultural exploitation. This activity is very important to Brazil because the history of Brazilian livestock has a lot to do with the country’s own history. Furthermore, it is some of the most important national agribusiness activity. Owning the world’s biggest commercial herd, the activity has placed Brazil among the top beef producers and exporters in the last years. However, the cattle breeders don’t have any professional business management which results in lack of management information. And also, this does not ensure the maximum possible income because the activity has a lot of productive and commercial risk. Forecasting beef prices is a way to minimize their commercialization risk. Therefore, the main objective of this essay is to compare multiple regression analysis and the ARIMA model of beef price forecasting by the methodological approach. This article suggests that the lack of management information of the cattle breeders really affect the business income. The only variable that could have an effect on beef price is the length of beef bided. Furthermore, the beef price is much more related to the price itself than to other variables of the market. Lastly, comparing the two forecasting models used it can not be said that one is higher or lower either in structure or in results. Since each model presents different attributes both of them were essential to the article. To future works, it is suggested that the beef price forecasting model use qualitative forecasting methods combined with quantitative within an artificial intelligence environment.