基于规则的分类数据离群挖掘方法研究

RULE-BASED OUTLIER MINING APPROACH IN CATEGORICAL DATA

离群数据的挖掘 (outlier mining,简称离群挖掘 )是数据挖掘的重要内容 ,现有的离群数据挖掘算法大多对分类数据 (categorical data)缺乏有效的处理 ,提出了基于规则的分类数据离群挖掘方法 ,采用多层最大离群支持度 maxsup,搜索离群规则 ,有效地解决了这一问题 ,用这一方法对医学流行病数据进行了各种实验 ,分析了该方法的适用范围、性能 ,验证了方法正确性 ;另外 ,实验表明 ,经过离散化后 ,基于规则的分类数据离群挖掘算法对连续性属性的数据也是有效的 .

Outlier mining is an important part of data mining. Existing outlier mining approaches lack valid processing for categorical data. A kind of rule based outlier mining approach in categorical data is introduced, which can solve the problem effectively. Some experiments have been done on epidemiology data in this method. The range of application and performance of this approach are analyzed, and it is found that if rule based outlier mining approach in categorical data is applicable in other continuous data after being discreted, good results can also be obtained.