Study On Application Of Data Mining Based On Bayesian Networks

Many methods have been used in data mining, Bayesian networks has become a focus currently. It has the broad application prospects. The main task of data mining is data analysis and processing, gains the implicit, prior unknown and useful knowledge. His ultimate purpose is to discover the characteristic concealing between internal law and data, thereby serving the management and decision-making. Bayesian networks is a new data modeling tool proposed at the end of last century. In the uncertainty inference carried on and the knowledge expressed, it has displayed its originality. When used in conjunction with statistical techniques, Bayesian networks has several advantages for data modeling.This paper devotes to Bayesian networks in data mining applied research. Firstly studies and summarizes the Bayesian networks correlation theories. The study of Bayesian networks is an extremely important link in data mining. This paper discusses the question of network structure, for the use of Bayesian networks to solve actual problems, the establishment of the data structure and the foundation for dependence. Secondly discusses data mining related knowledge as well as the mainstream algorithms, and analyzes each kind of algorithms good and bad points. The discussion proposes to use the data in data mining to construct a heuristic algorithm thought of Bayesian networks. It solves the algorithmic question in data mining using the sample data well. Finally utilizes this method to establish the university student to take exams for postgraduate schools model and a farmer credit scoring model, and separately carries on the comparison with the decision tree method and the traditional credit marking method. The experimentation indicates the methods proposed practicality, availability and high precision, confirms the superiority in data mining, helps our management, analysis, forecast and decision-making and so on.