To med customers' request, to manage their productions and their stocks or to direct their markding policies, textile companies must improve their supply chain management. This organization requires sales forecasting systems adapted to the uncertain environment of the textile field. The uncertainty is characterized bynoisy data and numerous explanatory variables (controlled, available or not) that influence the sales behaviour. This way, some recent researches introduced new forecasting models based on "soft computing". This paper deals with a comparison between new mean-term forecasting models and some traditional ones. The proposed models are hybrid neural model (HNCCX) and hybrid fuzzy model (HFCCX). They use neural networks and fuzzy logic abilities to map the non-linear influences of explanatory variables and consider the seasonality factors. The latter improvement, which is presented here, allows the reduction of models complexity by reducing the number of parameters. These models are also well -adapted to short and discontinuous time series, i.e. when the product sales occur during only some periods a year (major cases of textile items). To evaluate performances, the comparative test has been applied to real data of textile items, selected from an important French ready-to-wear distributor. Extensions are also proposed.
Textile-apparel industry, sales forecasting. artificial neural network, fuzzy inference system, production and distribution management, explanatory variables
All articles published by Studies in Informatics and Control (SIC) are published in accordance with the terms and conditions of the Creative Commons Attribution License (CC BY-NC-SA 4.0).