Application And Research On Beyas Classification Algorithm

This article focused on data mining method of Bayesian methods,it has a solid theoretical foundation of mathematics,and it can comprehensive informatian and data for prior information.therefor it becomes one of the hot spots on data mining and machine learning.Naive Bayes classifier is a simple and effective classification method on probability theory,but its attribute independence assumption is ofen violated in the world.To improve the performance of Bayes classifiers,in recent years,a great many of research has been done on constructing models which can express dependence among attributes.Some basic theories used in this thesis are briefly introduced, including Naive Bayesian classification modeling, Correlated Bayes and Weighted Naive Bayes, the basic thories of Rough Set . Based on the theory of rough set,a new Naive Bayes method named Mutual Information-based Algorithm for Weighted Naive Bayes (WCB) was proposed,which synchronously loosen Naive Bayes classifier's independence and equal importance of the attribute assumptions. Compared with Correlated Bayes (CB) and Weighted Naive Bayes (WNB),Simulation results on a variety of UCI data sets illustrate the efficiency of this method.Last, we summarize the research of this thesis and put forward some suggestions about further study.