基于空问定位的聚类算法在电信业客户划分中的应用

随着电信市场竞争的不断加剧,电信企业传统的营销模式正在向主动、精确营销模式转变.因此,客户划分显得至关重要.采用数据挖掘中聚类的方法对电信业的客户进行划分,在比较了现有聚类算法计算复杂度普遍较高的的基础上,采用了一种基于空间定位的方法,将客户数据对象映射到特征空间中,并利用空间立方体的某些特殊顶点定位任意数据,通过计算数据点与空间立方体顶点群的距离差异,完成聚类过程.为了适合电信业客户的特殊性质,改进对客户属性数据的处理过程.通过电信客户的数据实验结果表明,算法的时间复杂度降至 O(N) 级别.

Customer Demarcation Using Clustering Method Based on Space Location

With the competitions among telecom industry prick up day by day,the traditional sell pattern has been changing to the active and smart way,so customer demarcation become more important than before.Customers of telecom industry with the way of "Clustering" are compartmentalized,which are a usual technology of data mining.After compare with the time-complexity of common clustering methods,a method named "clustering based on space location" is choosed to use.First the object of customer date into special space is mapped,then every data-element using some special peaks of the cube space is located.At last,difference of distance between the date-elements and the special peaks of the cube space to finish the process of clustering is used.To fit the particularity of telecom customers,the way of customer date procession is improved.According to the result of experimentation using the customers data,the time-complexity has fall to 0(N)is found.