条件贝叶斯网络分类器及其在产品故障率等级分类中的应用

Conditional Bayesian network classifier and its application in product failure rate grade indentifying

针对传统贝叶斯网络分类器模型的不足,提出了一种基于条件贝叶斯网络的分类器模型。通过分析贝叶斯网络模型给定目标变量时各特征变量间的条件独立关系,充分利用其关联关系,为解决分类问题提供了一条有效途径。在此基础上,提出了基于条件贝叶斯网络分类器模型的建模方法用于指导实际模型建立和应用。实例分析结果表明,条件贝叶斯网络与其他的贝叶斯网络分类器及传统的决策树C4.5分类器相比,在提高分类器分类精度的同时降低了网络模型结构复杂度。

Aiming at the weakness of traditional Bayesian network classifiers,a new kind of classifaier model based on Conditional Bayesian Networks （CBN） was proposed. With the indication of the conditional independence relationship among attribute variables given the target variable,this model provided an effective approach for classification problems. Based on this,the modeling method for building CBN classifier was listed to guiding the modeling and application. Case study was carried out and the results showed that,comparing to existing Bayesian networks classifiers and traditional decision tree C4.5,the CBN not only enhanced the total precision but also reduced the complexity of network structure.