CA-GIS Model for Dynamic Simulation of Commercial Activity Development by the Combination of ANN and Bayesian Probability

Applying Artificial Intelligence (AI) techniques with Geographic Information System (GIS) in urban-related research is a surging trend recently. Artificial Neural Network (ANN) and Cellular Automata (CA) are two AI techniques discussed in this paper. ANN is known for its perceptron logic and CA is known for its dynamic evolvement mechanism. This paper offers the combination of ANN and Bayesian probability- named ‘BAANN’ process in order to integrate the advantages of ANN and CA. BAANN can transfer ANN outcome into the transition probability of target pattern, and cooperate with CA for operating dynamic simulation based on the perceptron logic. A specific experimental material is a kind of urban commercial phenomenon, which is the development of night market. The experimental CA model is implemented on GIS- named ‘CA-GIS’ model. CA-GIS model based on BAANN can perform high accuracy simulation and comprehensively evaluate land commercial value of night market. Additionally, the experiment of dynamic simulation can find out which places are the originally developing spots, and which zones would be sensitive to the situation of economic fluctuation.