Pulse graph characteristics analysis and classification of sub health state

Objective: To study on sub health state pulse graph classification characteristics and data mining classification method. Method: 1275 cases were divided into health and sub-health groups through health assessment by “health assessment questionnaire”(H20.V2009), another 121 disease cases in the control group; classifying the sub-health pulse graph characteristics by using naïve Bayes, support vector machine, decision tree, neural network data mining algorithm methods according to the pulse diagram parameter evaluation. Result: Decision tree algorithm total classification results on health pulse graph for62%, on sub health pulse graph total classification results for81.1%, on disease pulse graph total classification result is 49.1%, the decision tree algorithm of pulse graph total classification results for the 72%, better than the other algorithms. Decision tree algorithm is more suitable for different health states of the pulse index classification research. Conclusion: Decision tree algorithm was the effect optimal method in data mining on health, sub-health, disease pulse graph classification, data mining method facilitated the classification of pulse graph health, sub health and disease.