基于证据理论的知识发现分类算法

决策树方法是一种重要的可完成分类任务的知识发现技术 ,目的是通过构造一个分类模型 ,把数据库中的元组映射到给定类别中的某一个。决策树分类算法效率高且应用广泛 ,但是不能处理在决策树的构建和分类过程中的不确定数据。针对决策树分类算法的局限 ,利用证据理论是对概率论的扩展 ,将置信函数与概率的上下值相联系 ,可用于不确定数据的表达这个有力工具 ,把决策树分类技术扩展到含有不确定数据的环境中 ,提出了 D- S决策树分类算法。实验结果表明 D- S决策树分类算法能有效的对不确定数据进行分类。

A Knowledge Discovery Classification Algorithm Based on Evidence Theory

The decision tree is an important technique for accomplishing the classification task of knowledge discovery.The target of the classification is to find out a classification model.The model can map a single record in database to a pre-assumed class.Despite its popularity and efficiency,the standard decision tree is unfit to cope with data pervaded with uncertainty both at the construction and at the classification phase.Dempster-Shafer evidence theory extends from the probability theory and can handle uncertain data.In order to overcome these limitations encountered in a standard decision tree,this paper extends the decision tree technique to an uncertain environment where the uncertain is represented by evidence theory.This D-S decision tree is a new classification method adapted to the uncertain data.