Previsão de preço do milho com a aplicação de um modelo autorregressivo

Abstract

The present work analyzes and projects the price of corn in Brazil, which is one of the world's largest producers of agricultural commodities and whose agriculture contributes significantly to the GDP. Furthermore, Brazilian corn production is extensive and has a diversified production chain, including feed, feed production, and ethanol. Moreover, the study was developed using time series to analyze and project corn prices to help producers maximize their efficiency by having more data to improve their marketing and production planning and increase efficiency in the profitability of capital allocation. The analysis of time series is important to understand trends, seasonal and cyclical patterns, and random fluctuations in variables, such as corn prices. There are several methods for analyzing time series, and in this work, the Dickey Fuller test was used to verify the viability of using the series to make price projections. It was necessary for the series to be stationary, which means having constant mean, variance, and autocorrelation over time. In this way, the values are statistically consistent and change around a constant mean. Otherwise, transformations in the variables are required to make the series stationary. The ARIMA (Autoregressive Integrated Moving Average) model was used to make predictions, with a portion of the data separated for training and another for testing. Therefore, the objective of this work is to apply techniques for estimation and forecasting of time series to analyze the price series of corn in Brazil. It has been demonstrated that executing projections with autoregressive models in corn prices is feasible and viable, following the appropriate protocols. Furthermore, it is concluded that analyzing time series of prices allows for a better understanding of price behavior and risk mitigation.