基于Canopy-KMeans组合聚类的机票代理人行为刻画研究  

Airline Agent Behavior Characterization Method Research Based on Canopy-KMeans Combinatorial Clustering

作　　者：王洪建 Wang Hongjian(Xiamen Airlines,Xiamen 361006)

机构地区：[1]厦门航空公司,厦门361006

出　　处：《现代计算机》2021年第32期69-74,共6页Modern Computer

摘　　要：随着C2B的不断发展,民航机票销售代理人已成为航空公司重要销售渠道,而代理人的不规范销售行为会严重影响航空公司的销售收入及声誉。针对这一情况,本文在分析国内代理人销售数据的基础上,采用累加求和、对比分析、市场占比等方法重组生成了代理人活跃度、市场份额、买入卖出偏好等3大类18小类特征属性,提出了一种基于Canopy-KMeans聚类算法的机票代理人行为刻画算法,实证分析结果验证了基于大数据分析代理人行为刻画算法的正确性和有效性。其分析方法和结论有助于航空公司有针对性地采用不同的渠道策略、规范机票销售市场行为,具备现实的指导意义。With the continuous development of C2B, the civil aviation agents have become the important distribution channel of airlines. The irregular distribution behaviors of agents have seriously affected the revenue and reputation of airlines. Based on the analysis about domestic sales data, firstly, this paper reorganizes and generates 3 categories and 18 sub categories of attributes such as agent activity, market share, buying and selling preference by using the methods of cumulative summation, comparative analysis and market share, then Canopy-KMeans clustering algorithm was presented to classify all agents. Finally, The empirical analysis results verify the correctness and effectiveness of the agent behavior characterization algorithm based on big data analysis. The analysis methods and conclusions have practical guiding significance for airlines to adopt different channel strategies and standardize ticket sales market behavior.

关 键 词：机票代理人 行为刻画 组合聚类 属性重组 

分 类 号：F562.6[经济管理—产业经济] TP311.13[自动化与计算机技术—计算机软件与理论] F274[自动化与计算机技术—计算机科学与技术]