An online sequential feed-forward network model for demand curve prediction

Dynamic pricing of products is an important revolution in the retail and manufacturing industries, driven in large part by the E-commerce. In real application, the demand curve, i.e., how demand responds to changes in price, is critical to establishing any dynamic pricing model, but precisely estimating a demand curve that vary over time is a formidable challenge. In this paper, we consider the buyers purchase decision is dependent on multiple factors such as product price, positive feedback rate and prices offered by other competing sellers, and the demand curve is a time-varying function related with these factors. Hence, we propose a sequential feed-forward network model to capture the demand information from the real-time data and online predict the dynamic demand curve. The main benefits of the proposed method are spurning the structure assumption that is always used in the existing prediction methods and having fast and high accuracy. Some simulation results in comparison with results carried out by related works show clearly the effectiveness of the proposed method. Copyright © 2013 Binary Information Press.