The logistic regression method based on the data mining process -The application of the sub-health classification and the analysis of effect factors
亚健康人群分类及其临床特征分析与评价——基于数据挖掘流程的Logistic回归方法的研究

Objective: this paper aims to analyze the survey data using logistic regression method based on data mining process to get the final classification and the clinic characteristic of the sub-health crowd. Method: the sub-health epidemiological data is analyzed firstly by the whole data understanding and then by selecting variables and finally by choosing the appropriate model. Thus, the classification equation and the clinic characteristic of sub -health are obtained. Results: Two logistic regression models are established in two ways, each of which is also tested using testing data set to reach the classification accuracy. And the results are satisfying which show that the main clinic characteristics are body fatigue, sleep difficulty, bad memory, work efficiency declining, mental blankness, irascibility, etc. Conclusion: This method is superior to the traditional logistic regression method in dealing with the case with many explanation variables, showing great advantage.