Type 2 Diabetes, Discriminant Analysis And Logistic Regression Analysis

In recent years, marked by chronic high blood sugar of Diabetes mellitus developed rapidly on a global scale. Moreover, there is a yearly upward trend, and it has become a human health hazard chronic disease after tumor and cardiovascular disease,especially in developing countries . The prevention of Diabetes mellitus imminent. While Diabetes mellitus is a recognized that it is not the only cause of a single disease, which is a compound causes of syndrome. If we can extracted from the main factors and prevented Diabetes mellitus from the standpoint of the main factors among the many factors. It will play a positive multiplier effect.Multivariate statistical analysis of Discriminant analysis and Logistic regression analysis play an important role in classifying an individual and have broad application. For instance ,predicting the success of new products or failure,confirming a person credit risk or predicting a person whether suffer from lung cancer or not and so on. Once defining classification criteria, researchers can choose the explanatory variable to predict or explain each individual of kind and can diagnose the consultation on the explanatory variable analysis. However, two methods have their similarities, but there also has differences. Discriminant analysis is a method that on the basis of the known classification we base on the sample data and follow a certain criteria and establish discriminant function, then calculate the discriminant critical value and classify samples of the unknown type .Logistic regression model actually is a generalized linear model, It is a statistical tool of analyzing continuous date,discrete data and Contingency table data. More importantly it can systematically evaluate the link of variables and estimate the effect between variables.The main results of the paper are that we analyze the cause of Diabetes mellitus using two methods, including Discriminant analysis and Logistic analysis, and show some determinately conclusions. Furthermore we compare the same point and the difference of the two methods, as well as the excellences and shortcomings.