Radon Potential Prediction Research Based On Naive Bayesian Algorithm

Nowadays the influence of environment on human health attaches more and more inportance to people and many investigations and researches have been launched on environment. There are some radioactive gases that associated with human life in the air, like Radon. Radon can be generated through the uranium decay.It is a radioactive gas that widely exists in rock, soil and air. And it is also a natural gas. Radon is harm to the health which has the potential to cause lung cancer. So studying radon risk comprehensively is necessary in areas of high radon levels.China is a vast country and has a large population. It would be a huge task if take radon measurements in all places. Add Naive Bayesian Classifier to radon potential prediction will effectively reduce the workload for it is easy and effective and has a stable classification efficiency and a solid theoretical foundatio.In this paper 56 effective radon measuring points data and its related parameters at Zhongshan city of Guangdong Province are in the research scope. First make a intensive study on Naive Bayesian algorithm (NB) and Naive Bayesian algorithm based on the correlation and coefficient (WNB-CC). Using MATLAB software to write codes and verify its correctness. Then grading the 56 effective radon measuring points data and its related parameters by step. Determine the grading standard of the lithology, soil radon exhalation rate, uranium content and soil radon concentration.At the same time compare the NB algorithm prediction probability to WNB-CC algorithm The WNB-CC algorithm prediction probability is generally higher than that of NB algorithm. Then the unsuccessful predicting points of the single point prediction of WNB-CC algorithm are analyzed.Finally the acceptability for unknown point prediction are analyzed. The acceptance criteria of unknown point prediction is summarized combining with single point prediction results of NB and WNB-CC algorithms. On this basis, some unknown points in Zhongshan city are predicted. The results are acceptable.In this paper, the radon prediction research based on Naive Bayesian is aimed at reducing a lot of radon field investigation work and also provide some reference for future generations.