Research on Agent-Based Dynamic Pricing Model and Algorithms
Wang Guang-xing · Journal of Northeastern University · 2002
The regression trees algorithms is an alternate function approximation method itcan forecast numerical value true and fleetly. A combined algorithm that base the Q-learning and regression trees algorithm was proposed. It can improve Q-learning algorithm′s shortcoming that need excessively long training times. How to make a price in a competitive marketplace that uses the Q-learning and regression trees algorithms,and how to analyze average profile rate and training time′s relationship in the market model that use numerical value analyze method was presented. The curve graph about average profile rate and training time′s relationship was provided.