基于互信息与层次聚类双重特征选择的改进朴素贝叶斯算法

根据朴素贝叶斯算法的特征条件独立假设,提出一种基于互信息和层次聚类双重特征选择的改进朴素贝叶斯算法。通过互信息方法剔除不相关的特征,然后依据欧氏距离将删减后的特征进行分层聚类,通过粒子群算法得到聚类簇的数量,最后将每个聚类簇中与类别互信息最高的特征合并为特征子集,并由朴素贝叶斯算法得到分类准确率。根据实验结果可知,该算法可以有效减少特征之间的相关性,提升算法的分类性能。

Improved Naive Bayes Algorithm Based on Dual Feature Selection of Mutual Information and Hierarchical Clustering

According to the assumption of feature condition independence of the naive Bayes algorithm,an improved naive Bayes algorithm based on dual feature selection of mutual information and hierarchical clustering is proposed.The irrelevant features are eliminated by the mutual information method,and the deleted features are hierarchically clustered according to the Euclidean distance,and the number of clusters is obtained by the particle swarm optimization algorithm.Finally,the feature with the highest mutual information with the category in each cluster is merged into feature subsets,and the classification accuracy is obtained by the Naive Bayes algorithm.According to the experimental results,the algorithm can effectively reduce the correlation between features and improve the classification performance of the algorithm.