The Research Of Na(?)ve Bayesian Classification Based On Genetic Algorithms

The classification is an important research branch in the data mining domain. It has obtained many amazing achievements. Owing to its highly efficient and highly precise calculation, as well as its strict theoretical foundation, Naive Bayesian Classifier has obtained widespread application. However, its condition independence assumption limits its real application. The genetic algorithms are one kind of auto-adapted global optimization probability search algorithms which form through simulating the heredity and evolution process of the biology in the natural environment. Its simple, all-purpose, steady character has made great achievements in the solution to difficult, complex problems, and can convergence to the global minimum.Based on the genetic algorithms, Naive Bayesian classification method was studied in this dissertation. The main work were as follows:First, the research background and the primary mission of data mining were outlined, and the definition, the method as well as the classification model appraisal standard of the classification question in the data mining domain were described.Second, the Naive Bayesian classification model, the general principle of the model, as well as some existing questions was described.Third, the basic ideas of the genetic algorithms were elaborated, and one kind of improvement genetic algorithms namely auto-adapted genetic algorithms was described.Last, by introducing the genetic algorithms to the Naive Bayesian classification research, a Naive Bayesian Classification algorithm based on genetic algorithms (G_NBC for short) was proposed in this article. In order to avoid the effect of the training sets' noise and the data scale causing the influence of feature reduction not to be too approximately ideal, and then effecting the classification influence, this algorithm generates certain attribute subsets of the training sets through the random attribute selection, and constructs the corresponding Naive Bayesian classifiers, and then optimizes the Bayesian classifiers by using genetic algorithms. The experiments at the end of this dissertation confirmed the validity of this algorithm.