基于数据挖掘的客户细分技术

Client segmentation based on data mining Algorithm

针对数据挖掘算法中的聚类算法在聚类不规格形状数据点分布的处理难题,对基于密度梯度的聚类算法进行了研究。通过分析数据样本及其周边的点密度变化情况,选择沿密度变化大的方向寻找不动点,从而获取原始聚类中心,再利用类间边界点的分布情况对小类进行合并。阐述了基于密度梯度的聚类算法以及应用此算法进行电信行业客户细分的方法、步骤和案例。

Aimed to solve difficult problems in clustering algorithm based on density gradient is presented. With an points with the maximum density are searched and taken as with irregularly distributed data set, a new clustering alyses o original f density of each point and its neighbors, centers of clusters. Then some little clusters are combined into larger clusters according to the distribution of border points between clusters. This clustering algorithm can be used in the study of client segmentation in the field of telecom. Detailed discussions are then focused on the proposed clustering algorithm and its applications in telecom clients segmentation including methods, steps and cases