Weighted Naive Bayes classification algorithm based on particle swarm optimization

Abstract

Naive Bayesian classification method is applied in many fields, but in the real world, its properties do not satisfy the assumption of independence. Harry Zhang and Shengli Sheng extended the naive Bayes into weighted naive Bayes. This paper presents a Weighted Naive Bayes Classification Algorithm Based on PSO (particle swarm optimization, which was first proposed by Kenney and Eberhart). This method makes use of automatic search function of PSO, while maintaining the integrity of each attribute of data. According to the characteristics of the data itself, this method improves the classification accuracy of Naive Bayes and avoids the loss of information. Through the experiment on UCI data sets, expected results were achieved. The experimental results showed that the method was feasible and effective.