Análise comparativa de previsão de demanda: modelo geral versus modelos segmentados

Abstract

The present study presents a comparative analysis between the use of a general model versus segmented models forecasting of demand for the company Apti Alimentos Ltda. This study is characterized as an applied quantitative research. The objective of the research was to identify the best forecast model considering the segmentation of product families based on the ABC curve and the product life cycle analysis. To do so, we initially sought to describe the current forecasting method used by the company. Subsequently, it was tested whether the transformation of units from boxes sold to volume in kilograms sold yielded better results. From this, five models of demand forecasting were tested and the prediction errors of each model were calculated. Then, the aggregation of products was performed according to the representativeness of the ABC curve and the life cycle stage, in order to identify if the segmented forecast models have lower error rates than the application of a general model. It was identified that the segmented model according to the life cycle presents smaller errors, therefore, for the families that are in phase of growth and of maturity we must use the additive model of decomposition of the time series and the model of moving average of order three months for families who are in the declining phase. With the implantation of the model the company can have benefits in the management of stocks of raw materials and finished products.