Aplicação de métodos de previsão de demanda em uma indústria de alimentos no Triângulo Mineiro

Abstract

Demand forecasting is a strategic tool essential for production planning, inventory control, and waste reduction. This article aims to apply and compare different quantitative demand forecasting methods in a food industry, focusing on the category of 200 g cream cheese containers. For the analysis, historical sales data from January 2022 to December 2024 were used. The methods were applied using the Minitab® statistical software, and each model was validated based on the analysis of its respective errors. Five models were analyzed: simple moving average, simple exponential smoothing, double exponential smoothing, additive Holt-Winters, and multiplicative Holt-Winters, using Mean Absolute Deviation (MAD) and Mean Absolute Percentage Error (MAPE) as performances metrics. The results showed that the multiplicative Holt-Winters method achieved the best performance, with the lowest error values (MAD of 23443 and MAPE of 5 %), standing out for its higher forecasting accuracy. The application of this model enables improvements in the company’s supply chain management, contributing to more efficient decision-making and optimization of production resources.