Adaptive strategies for predicting bidding prices in supply chain management

Supply Chain Management (SCM) involves a number of interrelated activities from negotiating with suppliers to competing for customer orders and scheduling the manufacturing process and delivery of goods. Decision support systems for SCM need to be able to cope in uncertain, complex and highly competitive environments. Supporting dynamic strategies is a major but unresolved issue in the area. In this paper we examine two different approaches to address the issue of predicting customer offer prices that could result in orders in the domain of supply chain management. The first approach is to model the competitors' behaviour and predict their bidding prices according to the evolved models. The second one is to predict the lowest order prices for products for a number of days in the future using the time series of these prices. A set of algorithms are implemented based on Genetic Programming and Neural Networks learning techniques. The algorithms are tested in the TAC SCM simulated environment and the results are compared in terms of accuracy of prediction and execution time. Both learning techniques showed the potential for predicting prices in competitive and dynamic environments. The proposed Neural Networks algorithms demonstrate slightly better performance when tested in the TAC SCM environment compared to the algorithms implemented using Genetic Programming learning technique. Copyright 2008 ACM.