Combinação de técnicas de previsão de demanda: aplicação em uma agroindústria de nutrição animal

The increasing competitiveness of the markets in which organizations operate has required them to increasingly use accurate forecasting methods to aid in decision making processes. The objective of the present work was to analyze the performance of individual techniques and forecast combinations in the process of forecasting the fish feed demand of a small agroindustry. In this way, the Box-Jenkins and Holt-Winters models were used, as well as the combination of both, in order to fit a more accurate model. The combination of predictions was based on the arithmetic mean, regression and minimum variance methods with uncorrelated individual forecast errors, resulting in 5 proposed forecast. With the results obtained, it was verified that the regression model was the one that presented the best performance and should be used by the managers of the company to obtain the forecast of the future demand.