Agent-based location choice simulation

The paper is an attempt to show that the ABM (agent based modeling) can be used to simulate location choice and the simulating results are consistent with classical theory and the real situations. Existing location methods can express the relationships between site selection result and location factors, but there still are some limites to express the dynamic relationship between them. Hence, a dynamic, stochastic and forecastable location model should be built which can introduce the customer´s behavior into the model and combine the macro pattern and micro spatial interaction. So the authors propose using ABM to simulate the location choice. The classical Hotelling process is adopted as an example which is essential of the spatial process and can reflect the competition well. The model has 3 kinds of agents, i.e., customer agents, seller agents and government agents and the model is realized by StarLogo tools. In the model it is assumed that prices are fixed. There are homogeneous sellers and buyers and the products have no different. The simulating results are consistent with the classical Hotelling theory and the real analysis. The conclusions can be got from the paper that autonomic, flexible, communicable, reactive and pro-active, planning, learning and adaptable ABM have advantages in location choice than existed methods. A further ambition is introduce the non-market factors such as price competition, relocation cost and policy restrict to the ABM to simulate the more complex location choice of real world.