Análise e previsão do preço da soja com métodos de regressão linear e suavização exponencial

Abstract

The present study aims to analyze and forecast the behavior of soybean prices in Brazil using two well-established statistical methods: linear regression and exponential smoothing. Soybeans are one of the country’s main agricultural commodities, playing a strategic role in both the Brazilian and global economies. Their price is strongly influenced by factors such as supply, demand, climate variations, and international market conditions. These fluctuations directly affect the profitability of the agricultural sector, making price forecasting an essential tool for efficient planning and decision-making by producers and investors. In this study, historical price data were used to develop predictive models capable of capturing trends and seasonal patterns inherent to the soybean time series. Linear regression was applied to estimate the long-term price trend, while the exponential smoothing method stood out for its effectiveness in capturing seasonal patterns, especially during periods of high volatility. The results demonstrate that combining these two methods provides more accurate and reliable forecasts, enabling agricultural sector stakeholders to make strategic decisions based on trustworthy results. Furthermore, this study reinforces the importance of using complementary approaches in time series analysis, particularly in volatile markets such as soybeans. Thus, this research contributes to the application of price analysis techniques in agribusiness, providing support both for producers seeking to optimize their commercialization strategies and for policymakers who require precise results to implement policies aimed at ensuring the sector’s economic security.