基于Map/Reduce的民航高价值旅客发现方法  

Method of discovering high-value passengers of civil aviation based on Map/Reduce

作　　者：曹卫东[1,2] 白亮[1] 聂笑盈[1] 

机构地区：[1]中国民航大学计算机科学与技术学院,天津300300 [2]中国民航信息技术科研基地,天津300300

出　　处：《计算机工程与设计》2015年第4期1078-1083,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(60879015);中国民航局科研基金项目(MHRD201130)

摘　　要：为解决常旅客计划模型评价指标单一,不能准确识别高价值旅客的问题,提出一种将Map/Reduce并行处理与数据挖掘知识相结合的发现方法。利用Map/Reduce数据处理模型,在Hadoop分布式平台上并行处理海量PNR数据;根据改进的RFD模型,确定客户价值指标,利用AHP层次分析法将专家经验值量化为指标权重;利用聚类分析技术识别高价值旅客,采用真实的PNR数据集进行实验。实验结果表明,该方法能够有效识别民航高价值旅客,为航空公司做出有效决策提供有利依据。To solve the problem that the evaluation index of frequent flyer program model was single,which can not accurately identify high-value passengers,a method of discovering high-value passengers combining Map/Reduce and data mining was presented.Gigabytes of PNR data were parallel processed on Hadoop using Map/Reduce.According to the improved RFD model and analytic hierarchy process,the customer value indexes and the weight of each index were determined.The high-value passengers were identified by data mining,and an experiment was carried out on a real PNR data set.The experimental result indicates that,this method can identify the high-value passengers of civil aviation effectively and provide a favorable basis for airlines to make effective decisions.

关 键 词：Map/Reduce映射归约 数据挖掘 RFD模型 AHP层次分析法 客户价值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]