Previsão de demanda: seleção de um modelo quantitativo para o grupo de produtos porcelanato em uma empresa fabricante de revestimentos cerâmicos

Abstract

Faced with a competitive market, it is essential that companies anticipate expected demand, understanding fluctuations and seasonality, since that, within their limits of actuation, it is possible to control levels of inputs and inventory, the advantage of informing deadlines, and always have a close relationship with their customers. In this way, this work uses the temporal series demand forecast calculation to select the quantitative demand forecast model that best suits the porcelain tile product group. The company under study is a manufacturer of ceramic tiles, including wet porcelain tiles and wet and dry ceramics. To perform the demand forecast, the sales history of the 62.5x62.5 cm2 porcelain tile format was used, between the years 2017 and 2019, and in possession of this database, the simple moving average, weighted moving average, simple exponential smoothing, Holt method and Holt-Winters method calculations were performed. After performing the simulations in Microsoft excel software, especially the Solver and Data Analysis functions, the smoothing parameters and/or weights were found for each forecast model, which promoted the lowest value of the mean squared error. As a result, the model that demonstrated the most satisfactory performance was the weighted moving average with distribution of the weights over the previous 12 months. Besides presenting themselves as relatively simple forecast models that can be replicated for other formats, no financial resources are required for their use, which makes the realization of a forecast model as a pilot favorable.