摘要
针对电信客户"亲友通话圈"定价决策的需要,提出一种基于有向图的蚁群聚类算法对电信客户进行聚类.该算法在构造客户通话有向图的基础上,利用蚂蚁在搜索过程中不断积累信息素,更新有向图,并通过划分强连通分量得到亲友通话圈.通过对真实数据集的测试,算法可以有效、快速地形成聚类,合理地划分亲友通话圈.算法可以针对若干不同的阈值产生不同的聚类结果,选取其中成本最小者,从而获得最大利润,有效解决了通话圈定价的问题.
To make sensible pricing decisions for the ＂relatives-friends calling circles＂ of telecommunication customers, this paper presents an ant colony clustering algorithm based on directed graph to cluster the telecommunication customers. By constructing the directed graph of customers＇ communications, the algorithm modifies the directed graph by accumulating pheromone in the process of ants＇ searching, and obtains the ＂relatives friends calling circles＂ by dividing the strongly connected components. Experimental results on real data set show that this algorithm can form clusters effectively and rapidly, classify the ＂relatives friends calling circles＂ reasonably. By comparing the clustering results obtained using different thresholds, the scheme with the least cost is selected so as to make the largest profit and effectively solve the problem of pricing decisions for the telecommunication customers calling circles.
出处
《扬州大学学报（自然科学版）》
CAS
CSCD
北大核心
2009年第3期62-65,共4页
Journal of Yangzhou University：Natural Science Edition
基金
国家自然科学基金资助项目(60673060
60773103)
江苏省自然科学基金资助项目(BK2008206)
江苏省高校自然科学基金资助项目(08KJB520012)
关键词
聚类
蚁群算法
电信
客户关系管理
通话圈
clustering
ant colony algorithm
telecommunication
customer relationship management
calling circle
