Application comparison of two clustering algorithms used in customer behavior analysis
两种聚类算法在顾客行为分析中的应用比较

This paper summarized the design ideas and features of two kind clustering algorithms, such as GN clustering of network community division and AP clustering in pattern recognition. To serve as an example of the customer group, it constructed the data set from library borrowing records, and then made the comparison of two clustering algorithms. The results indicate that the two kind algorithms have revealed much more about the structure of the customer group, the outcome of GN clustering algorithm is close to customer macrostructure, and the result of AP clustering algorithm reflects the customer requirement distribution. The effect of algorithm design principles on experiment results has been discussed. This work can give a valuable reference for design improvement of clustering algorithm and customer behavior data mining.