Abstract: With the acquisition of customers service and ontological attribute data, the customer group's common features are analyzed by using multi-dimensional normal cloud model. Hence, a customer group's characteristic matrix is built. Then, by means of similarity measures based on cloud model,similarity analysis between customer groups is carried out in a qualitative way such that the customer classification can be dynamically adjusted. After determining the typical customers in each customer group, it can generate the activated intensity of individual customer's property value with respect to customer group's property value by using multi-dimensional X condition cloud generator. Then, the difference between individual customers and the typical customer in a customer group can be calculated. Numerical simulation shows that the multi-dimensional cloud model is useful for customer behavior analysis.

Key words: customer behavior, customer base, N-dimensional cloud model