Research On The Approach Of Classification In Data Mining Based On Naive Bayesian

Classification is one of the most important research topics in data mining. Its goal is to designate the most suitable category labels for some instances described by attributes sets. There are many classified methods and technologies for the construction of classified models. Bayesian method becomes a hot spot in data mining research, because it has a solid theoretical foundation in mathematics and the capacity to integrate priori information and data samples information. Classification based on Bayesian theory has two main branches: they are Naive Bayesian and Bayesian network.Among some classified methods, Naive Bayesian Classifier has been a focus of research because of its simple algorithm and effective calculation. Though the traditional Naive Bayesian Classifier has strong assumption, that is, all attributes are mutually independent, it is not true in reality. Such a case, to some extent, affects the performance of the classifier.Through studying several Bayesian classifiers, we analyzed their individual characteristics. In order to take advantage of the surperiority of Naive Bayesian Classifier (NBC) and its classified effect in classification, we suggest some proposals for improving its strict limitations of independent assumption and achieved some good results.The first chapter is a preface, which introduces the main concepts about data mining and knowledge discovery, including the process of data mining and the data mining' s functions. We introduce the current development and future trends of data mining, and explain the definition, methods of classification problems and the standard evaluation of classification models in detail.In the second chapter, we introduce the classified technology based on Bayesian in summary. We firstly introduce the basic knowledge of Bayesian theory. Secondly we introduce a few Bayesian classification models, such as Naive Bayesian Classifier, Bayesian Network Classifier and Incremental Bayesian Classifier. We analyze their individual characteristics in order to extend our concepts of Bayesian classification.In the third chapter, we introduce the basic theory of rough sets. Firstly, we introduce the basic concept of rough sets, knowledge reduction and the relativity of knowledge. Secondly, we introduce the research work of rough sets in theory and application. Then we discuss the effective algorithms about rough sets.In the fourth chapter, having compared several expanded models of Bayesian, we discuss how to improve naive Bayesian classification and propose one classified Bayesian model. Through searching attributes space,...