Abstract: An improved naive classification algorithm is presented to solve the three problems that affect the accuracy of naive Bayes algorithm. The gray related degree about condition attributes and classes is calculated, according to which the attribute joining and attribute weighted are completed. The absent attributes are filled with an improved EM algorithm. The samples are classified by Bayesian classification algorithm. The results of experiments indicate that the optimization algorithm has the higher efficiency for clustering.

Key words: grey relational degree, condition attribute, class attribute, attribute joining, attribute weighted

摘要： 针对朴素贝叶斯算法存在的三方面约束和限制，提出一种数据缺失条件下的贝叶斯优化算法。该算法计算任两个属性的灰色相关度，根据灰色相关度完成相关属性的联合、冗余属性的删除和属性加权；根据灰色相关度执行改进EM算法完成缺失数据的填补，对经过处理的数据集用朴素贝叶斯算法进行分类。实验结果验证了该优化算法的有效性。

关键词: 灰色相关度, 条件属性, 类别属性, 属性联合, 属性加权