Research on Online Shopper Behavior Pattern Based on a Dynamic Architecture Neural Networks and Genetic Algorithm

With the development of Internet, electronic commerce develops rapidly and grows strong. Today, it is very important for consumers to buy products by online shopping. Particularly, more and more consumers do shopping by mobile. However, the many researches indicate consumers often access the websites for shopping, and they give up their purchase viewpoint in the end. What factors influence consumer’s willingness to buy? To solve the problem, a novel and efficient algorithm is presented in this paper. The algorithm is excellent in modeling dynamic architecture by neural networks. In addition, a data mining algorithm which extracts relational rules from data records is proposed too. Genetic algorithm is used in this mining algorithm. Most of traditionary neural network architectures are arbitrary. Before train of neural network, firstly, the algorithm fixes the number for hidden layer and node. However, in this new algorithm, the numbers are dynamically produced. Moreover, the paper presented a novel genetic algorithm whose computation is simple. The optimal values of chromosome (input attributes), xm, is found by use of the genetic algorithm. The achieved results confirm the presented algorithms are efficient and novel.