Sales forecasts in clothing industry: The key success factor of the supply chain management

Abstract

Like many others, Textile–apparel companies have to deal with a very competitive environment and have to manage consumers which become more demanding. Thus, to stay competitive, companies rely on sophisticated information systems and logistic skills, and especially accurate and reliable forecasting systems. However, forecasters have to deal with some singular constraints of the textile–apparel market such as for instance the volatile demand, the strong seasonality of sales, the wide number of items with short life cycle or the lack of historical data. To respond to these constraints, companies have implemented specific forecasting systems often simple but robust. After the study of existing practices in the clothing industry, we propose different forecasting models which perform more accurate and more reliable sales forecasts. These models rely on advanced methods such as fuzzy logic, neural networks and data mining. In order to evaluate the benefits of these methods for the supply chain and more especially for the reduction of the bullwhip effect, a simulation based on real data of sourcing and forecasting processes is performed and analyzed.