Sales Forecasting through Fuzzy Neural Networks

In business strategy, sales forecasting is increasingly getting more and more attention. Numerous academic studies addressing this problem have generally employed statistical methods, such as regression or autoregressive and moving average (ARMA). However, sales forecasting is rather complicated because of numerous internal and external factors comes from the surrounding environment. Artificial neural networks (ANN) have been introduced into the sales forecasting method as their expected performance in the field of control. Even though, because of particular industry features, such as, promotion, advertisement, style and life span, all these determinants will influence the performance of ANN, so this study applies the Fuzzy neural networks (FNN) which is constructed of computational intelligence that come with significant learning abilities and inherent transparency. The real-world problem results not only show that FNN with weight elimination have lower training error, but also the outperformed performance of this proposed method compared with traditional statistical method and single ANN in accuracy.