基于SOM&SVM组合分类器的客户细分方法实证研究  

Empirical Research of Customer Segmentation Methods Based on Som&Svm Combining Classifiers

作　　者：邰丽君[1] 胡如夫[1] 赵韩[2] 陈曹维[1] 

机构地区：[1]宁波工程学院机械工程学院,浙江宁波315000 [2]合肥工业大学机械与汽车工程学院,安徽合肥230009

出　　处：《数学的实践与认识》2012年第11期139-146,共8页Mathematics in Practice and Theory

摘　　要：为解决传统的RFM客户细分方法还不能很好地刻画客户行为,同时也没有就RFM指标权重进行分析这一问题,在RFM指标的基础上扩充了客户细分的指标体系,并提出了基于AHP的RFM指标权重确定策略.鉴于传统的单一分类器存在的很多缺陷,提出基于SOM&SVM的组合分类器模型,充分利用SOM和SVM单一分类器各自的优点,综合两种分类器的分类信息,避免单一分类器可能存在的片面性,从而提高分类的准确性.最后通过实例对上述模型的有效性进行验证.To solve the problem that the traditional RFM customer segmentation methods Call not describe customer behavior preferably and did not analysis the RFM index weight, the customer segmentation index system was expanded on.the basis of RFM index and the RFM index weight determination strategy based on AHP was proposed. In view of a great many defects exist in the traditional single classifier, a combining classifiers model based oil SOM＆SVM was proposed, the respective advantages of SOM＆SVM single classifier was fully used, the classified information of two classifiers were synthesized, to avoid one-sidedness may possibly exist ill the single classifier, in order to improve the classified accuracy. Finally , effectiveness of the proposed model was Validated by an example.

关 键 词：SOM SVM 组合分类器 客户细分 

分 类 号：F274[经济管理—企业管理] O225[经济管理—国民经济]