«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

期数:

2006年03期

页码:

1-4

栏目:

智能、算法、系统工程

出版日期:

1900-01-01

文章信息/Info

Title:

Mining Association Rules in Customer Behavior in Telecommunication

文章编号:

1005-3751（2006）03-0001-04

作者:

梁循¹; 2

[1]北京大学计算机科学与技术研究所[2]斯坦福大学管理科学系

Author(s):

LIANG Xun

[1] institute of Computer Science and Technology, Peking University[2]Department of Management Science, Stanford University

关键词:

关联挖掘; 聚类; 连通子图; 客户行为

Keywords:

mining; soeiation rules;  clustering;  connected subgraph;  customer behavior

分类号:

TP391

文献标志码:

A

摘要:

提出了一种基于关联规则挖掘的聚类方法。首先，通讯行业客户行为的原始数据经过数据预处理转变为地区间的“距离”数据。其次，由于地区是“漂浮”的，不再是“刚体”，而是一种抽象的“柔性”距离，使用关联规则进行挖掘成为一种好的选择。文中对通讯行业客户行为进行了基于关联规则的建模，较好地嵌入了关联规则的框架。在数据实验后，提炼出了知识，发现东南亚客户聚成一类，以此为模式，得出了“在南美发展业务是错误的”的结论，该结论在挖掘之前是没有意料到的。实践上，该结论阻止了相应公司的南美发展计划，为公司度过后来的硅谷经济萧条时期省下了上百万美元的“战略储备”资金

Abstract:

Presents a method of clustering based on mining association rules. First, the raw data for the customer behavior in telecommuni- cation are transformed into the ＂distances＂ between the areas. Second, since the areas are ＂floating＂, as opposed to ＂rigid＂, the application of the association rule technique to the ＂flexible＂ distances turns out to be an adequate option. The customer behavior in telecommunication is calibrated into the framework of association rules. Experiments hint us the pattern by grouping the customers into a cluster. Based on the pattern, an analysis results in a conclusion that ＂it is a wrung decision to develop the business in South America＂, which is unexpected before data mining. In practice, the conclusion was applied and prevented the plan of South America, saving some one million US dollars of the strategic fund for living through the years of stagnant economy in Silicon Valley for the corporation

备注/Memo

备注/Memo:

留学回国启动基金资助项目（4131522）;国家自然科学基金资助项目（70571003）梁循（1965-），男，北京人，博士，博士后，MBA副教授，研究方向为电子金融、数据挖掘

常用功能

更新日期/Last Update: 1900-01-01