Metodologia para previsão de demanda em uma indústria com características de customização em massa

Abstract

Nowadays, the goods and services customization, introduced through the Mass Customization concepts, has become an important industry strategy in the search for opportunities to remain competitive in the market. That strategy implies a new reality regarding the challenges for Companies Operation Management whereas it becomes necessary to develop new business management methodologies. This Master Dissertation proposes a demand forecast methodology for seasonal customized items. The model considers a scenario where there is an ample offer of combinations for customized items and a seasonal environment. It is proposed the combination between of correlation clustering method and ARIMA model to forecast demand. The accuracy of the proposed forecast method performed is better than the current company method in clusters that represent 85% of sales. The MAPE resulted from the proposed demand forecast methodology was 33.6% in comparison of 69.7% obtained from company methodology. That results indicate a possibility of better adjustment in the items set that must be produced for inventory when the market demand does not reach the company production capacity level, guaranteeing cost and delivery time, characteristics of Mass customization environments to be reached because it allows a bigger number of items to be considered in the forecast.