摘要
针对彩铃业务交易记录较多和客户属性的高维度及混合性的特点,建立了基于信息熵度量的模糊粗集属性约简和RBF-SVM分类的彩铃客户挖掘模型。通过10折交叉验证,对来自两个地市的营销返回样本,在选择特征数量和分类精度之间的差别与其他5个模型进行了比较分析。实验结果显示此模型获取了相对最高的平均分类精度(80.43%)和最少的平均特征属性(2.5个),有效地约简了属性并改善了分类能力。
Aimming at the more dealing track record of the color ring operation, and the high-dimensional and hybrid properties of the customers attribute, based on the attribute reduction of fuzzy rough set with information entropy measure, and the RBF-SVM classifier, a new mining model of the color ring customers is built. Combining with the 10-fold cross validation, for the marketing data sets of two regions, the number and accuracy of selected features are compared with the other five models. Experimental results show that this model can obtain the relative and much better average classification accuracy （80.43%）, and select the least average feature attribute（2.5）, effectively reduce attribute, even improve classification power.
出处
《计算机工程与应用》
CSCD
2013年第4期125-128,共4页
Computer Engineering and Applications
基金
黑龙江省教育厅人文社会科学项目(No.12514128)
关键词
信息熵
模糊粗糙集
支持向量机
彩铃客户
information entropy
fuzzy rough set
rbf-svm
color ring customers
