两种聚类算法在顾客行为分析中的应用比较

分析网络群落划分的GN聚类和模式识别中AP聚类两种算法的设计思想和特点;以图书借阅记录为例构建了顾客聚类的数据集,进行了两种算法的聚类比较。研究表明,两种算法从不同角度揭示了顾客群体的结构特征,GN聚类结果与顾客的宏观特征分类相接近,而AP算法结果反映出顾客需求的分布特征。探讨了算法设计原则对实验结果产生的影响。这些工作可为聚类算法的设计改进和顾客行为的数据挖掘等研究提供一定的参考。

Application comparison of two clustering algorithms used in customer behavior analysis

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.