The Preliminary Application Of Bayesian Classifier On Risk Assessment Of Radon

More than 55% harm brought by environment radiation is from radon which is the only natural radioactive gas. It is the most leading cause of lung cancer except smoking. Because of its very uneven distribution in nature, it is important to assess the radon level for being targeted to take some effective measures to reduce and control the hazards of radon gas to the human.Bayesian classifier has low time complexity, high accuracy, and a solid theoretical foundation, has been widely used. The main point of the research paper is application of Bayesian classifier on risk assessment of radon. The factors that affect on concentration of radon were used for hazards assessment of radon as the condition attributes.The work for writing this paper:(1)Studying The basic principles of Naive Bayesian classifier,learning the applications of Naive Bayesian classifier, and trying to be applied to risk assessment of radon.(2)Programming for calculation of Naive Bayesian classifier, and verifying it’s correctness.(3)Collecting the relative data in some place, including radon data, geologic data, soils data and so on.(4) Performing the experiments of Naive Bayesian classifier based on the data from southern coastal of china and New York State of U.S.The results of the data experiments were acceptable, so it is feasibility to the application of Naive Bayesian classifier in risk assessment of radon. And then the conclusions were reached:(1) It is important to select the condition attributes, because they affect the accuracy directly. Because of the lack of a condition attribute, the accuracy is not as good as it might be. On the other side, if less precision required, it is more efficiency to give up the condition attributes which is difficult to collect. (2) Naive Bayesian classifier is no less than its development. Requirement of Naive Bayesian classifier condition attributes are independent, but the factors that affect on concentration of radon are less correlative. So The efforts toward the development of Naive Bayesian classifier are ineffectual—little the accuracy improving, slower the calculation speed.