基于贝叶斯方法的决策树分类算法  

Decision tree classification algorithm based on Bayesian method

作　　者：樊建聪[1] 张问银[2] 梁永全[1] 

机构地区：[1]山东科技大学信息科学与工程学院,山东青岛266510 [2]临沂师范学院计算机系,山东临沂276005

出　　处：《计算机应用》2005年第12期2882-2884,共3页journal of Computer Applications

摘　　要：针对数据挖掘的特点和本质,充分利用贝叶斯方法和决策树分类的优点,将贝叶斯的先验信息方法与决策树分类的信息增益方法相结合,提出了一种新的数据挖掘分类算法(BD1.0算法),并对此算法进行了设计和分析。实验分析表明,该算法可以处理不一致或者不完整数据等“脏数据”,比单纯使用贝叶斯方法或决策树方法具有更高的准确率,而且与C4.5算法具有近似的时间复杂度。According to the characteristic and essence of data mining and taking advantage of Bayesian method, a new classification method named BD1.0 algorithm was presented. This method combined the prior information and information gain method of decision tree, The design and analysis of the algorithm was introduced too, The experiment results show that the algorithm can deal with dirty data such as incomplete data or inconsistent data, and it is more accurate than only useing Bayesian method or decision tree, It has approximate time complexity with C4.5 algorithm,

关 键 词：数据挖掘 分类 贝叶斯原理 决策树 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]