摘要： 针对民航客户行为数据的复杂性，运用数据挖掘中的DBSCAN聚类技术，结合核映射机理，提出一种基于核的DBSCAN算法，用于实现民航客户的细分。实验结果表明，该方法能突出客户之间的行为特征差异，降低聚类结果的混乱性，且其聚类纯度比原DBSCAN算法约提升30%。

关键词: 客户关系管理, 数据挖掘, 核DBSCAN算法, 民航客户细分

Abstract: This paper proposes a kernel-based DBSCAN algorithm which is aiming at the complexity of the civil aviation passenger behavior. The algorithm uses the DBSCAN clustering technology, combines with the nuclear mapping mechanism, and realizes civil aviation customer segmentation. Experimental results show that the method can highlight the differences between samples, and also can reduce the confusion of the clustering results and the cluster purity is improved by nearly 30%.

Key words: Customer Relationship Management(CRM), data mining, kernel DBSCAN algorithm, civil aviation customer segmentation

中图分类号: 

• TP391