基于属性分类和回归模式的知识发现研究

Knowledge Discovery Studying Based on Attribute Classified and Lin ear Regress Pattern

该文针对目前数据挖掘的研究状况,理论上提出了将基于属性分类方法和多元线形回归算法相结合的算法,首先使用基于属性分类的方法将原始数据库进行属性分类,化简,去掉次要的条件属性,最后得出一个简化的表格,找出影响决策属性的主要因素,根据此表,可以得出简单的ifthen规则;然后使用多元线形回归求出它们之间的近似定量关系,得出一个最优回归方程。

Considering the present condition of Data Mining study,This paper presents an approach combining the method based on attribute classified and the method of Poly-factor Linear Regress Analysis.First,it uses the method based on attribute classified to classify and simplify attributes in raw database,weeds out the secondary condition attribute,gives a result of simplified form,discoveries the main factors influencing decision attribute.According to this form,you can get simple″if-then″rules.Then it uses the method of Poly-factor Linear Regress Analysis to get the implied approximately relationship between them.At last,it presents a best regress equation.