贝叶斯分类的研究及应用

Research and Application of Bayesian Classification

摘要: 本文研究的是贝叶斯分类方法的研究现状和在实际中的应用。文章的前半部分对贝叶斯分类的研究现状进行了介绍。文章后半部分对主成分分析法进行了介绍和实现，并使用碎石图将主成分分析的结果可视化。然后将主成分分析与加权属性结合处理后的数据与贝叶斯分类器结合，给出了一个改进的基于主成分分析的加权贝叶斯分类方法；其次通过互信息求出特征词与特征词所在类别的概率，将此概率作为贝叶斯分类器的先验概率进行分类，给出了一个互信息特征选择的贝叶斯分类算法。最后，经过数值实验验证了提出的两种改进方法都有着比较好的分类效果。

Abstract: This paper studies the research status of Bayesian classification and its application in practice. In the first half of this paper, the research status of Bayesian classification is introduced. In the last part of the paper, the principal component analysis method is introduced and realized, and the results of principal component analysis are visualized by using the rubble diagram. Then, an im-proved weighted Bayesian classification method based on principal component analysis is proposed by combining the processed data with weighted attributes and Bayesian classifier. Secondly, the probability of feature and category of feature is calculated by mutual information, which is used as the prior probability of Bayesian classifier for classification, and a Bayesian classification algorithm for feature selection of mutual information is presented. Finally, numerical experiments show that the proposed two improved methods have good classification performance.