Previsão de demanda como estratégia para melhoraria do abastecimento de água mineral: estudo de caso em uma empresa do varejo farmacêutico

Abstract

This study aims to evaluate the adequacy of demand forecasting methods applied to the commercialization of mineral water in a pharmacy retail chain in the state of Ceará, considering the direct supply model adopted by the company. It is an applied research study with a quantitative approach, employing bibliographic research procedures and a case study. The analysis was based on historical sales data of bottled mineral water, covering the period from January 2022 to November 2025, with emphasis on stores representative of demand behavior. Initially, the performance of demand forecasts using the simple moving average method, currently adopted by the company, was evaluated. Its limitations include high reactivity and the inability to anticipate seasonal variations, resulting in stockouts and excess inventory. Subsequently, seasonal adjustment and Holt–Winters methods were applied and compared, as they are suitable for explicitly addressing trend and seasonality components. Model performance was assessed using two error metrics: Mean Absolute Deviation (MAD) and Mean Absolute Percentage Error (MAPE), considering both training and validation datasets. The results showed that the Holt–Winters method achieved higher predictive accuracy for both analyzed stores, demonstrating greater robustness in contexts of high consumption variability. It is concluded that replacing the current method with approaches more appropriate to the stores’ demand profile can significantly contribute to improving demand planning and inventory management within the company