摘要
针对目前航空公司旅客细分工作不够精细的问题,在分析传统RFM模型的基础上,提出一种TCSDG模型来描述旅客行为偏好。根据旅客的行为偏好对旅客进行细分,将具有相同行为偏好的旅客聚成一簇,为航空公司针对不同行为偏好的旅客提供个性化服务提供基础;结合Hadoop并行化计算平台,将算法并行化,以处理海量订票数据。在中国民航订座系统数据上的实验结果表明,该算法在保证细分结果的基础上提高了执行速度和处理能力,根据旅客行为偏好高效地将旅客分为不同的簇,使行为偏好相同的旅客聚成一簇。
To solve the problem that the passenger segmentation of airline is rough, a TCSDG model based on RFM model was proposed. The preferences of passengers were obtained and the same preferences of passengers were clustered together, so that airlines provided personalized recommendation service to the different clusters. The algorithm was parallelized to deal with mas- sive amounts of booking data by combining the Hadoop parallel computing platform. Results of experiments on real booking data show that the algorithm can classify passengers into different clusters according to different behavior preferences efficiently.
出处
《计算机工程与设计》
北大核心
2015年第8期2217-2222,共6页
Computer Engineering and Design
基金
国家自然科学基金项目(61139002)
中国民用航空局科技基金项目(MHRD201130)
中央高校科研业务经费基金项目(3122013C005)
关键词
客户关系管理
客户细分
行为偏好
并行聚类
TCSDG模型
customer relationship management （CRM）
customer segmentation
behavioral biases
parallel clustering
TCSDG model
