摘要
数据挖掘技术为高效的客户分类提供了强大的支持,然而仅依靠这门技术并不能很好地完成这项任务。因为分类方法的局限性,现实数据存在信息的不确定、不完整、先验知识缺乏,研究对象的复杂性等困难导致的分类不确定性。从这个角度出发,将模糊积分融合方法与数据挖掘技术结合来减小客户分类的不确定性,提出了一种模糊密度修正方法,它利用了训练样本先验静态信息和各分类器识别结果包含的动态信息对模糊密度进行自适应动态赋值。仿真结果表明了它的有效性。
Though data mining technique provides powerful support to highly efficient client classification,the technique can not fulfill the task well alone.Becanse of the limitation of this classifying method,the classifying uncertainty,which is resulted in the difficulties such as the uncertainty,the incompleteness and the deficiency of transcendent knowledge of the information and the complexity of the research object,exists in the real data.From this point of view,the uncertainty of client classification is decreased by combining the method of fuzzy integral and the technique of data mining and a blur density correcting method is put forward to automatically adapt and dynamically evaluate the blur density by using transcendent static information of the training prototype and dynamic information included in the identified result of various classifiers.The emulational result testifies its validity.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第1期212-214,248,共4页
Computer Engineering and Applications
基金
国家自然科学基金重点资助项目(60234030)。
关键词
信息融合
模糊积分
不确定性
information fusion
fuzzy integral
uncertainty
