摘要
针对客户市场细分问题进行了研究。依据粗糙集理论,以信息表中条件属性与决策属性的一致性原理为基础,以超立方体、扫描向量为数据计算对象,进行数据离散化和连续型属性约简,实现了数据预处理;在此基础上,以集合差异度指标为约束条件,运用集合特征向量加法法则最终实现客户市场细分。在实验中,连续属性离散化和冗余属性约简有效地减少了计算数据,便于客户市场细分的实现,提高了客户市场细分的效果。研究结果表明该客户市场细分算法是有效可行的。
This paper deals with the problem of customer market subdivision.Based on rough set theory,depending on the consistency of condition attributes and decision attributes in the information table,by taking the data super-cube and the scan vector as the calculating objects,continuous attributes were discretized and redundant attributes were reduced.Customer market subdivision was then implemented through the addition rule of set feature vector with the constraints of set dissimilar degree.In experiment,many repeated data were deleted and redundant attributes were reduced in the process of discretization and attributes reduction.Experimental results show that the algorithm for customer market subdivision is efficient and effective.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第z1期1064-1068,共5页
Journal of Tsinghua University(Science and Technology)
基金
中国博士后科学基金资助项目(2005038319)
辽宁省教育厅重点科技基金资助项目(202163345)
关键词
粗糙集
扫描向量
属性约简
离散化
rough set
scan vector
attributes reduction
discretization
