Using Double Logistic Regression To Improve Discriminant Efficiency

Linear regression model is widely used in quantitative analytical techniques, it is used to predict or explain dependent variable by independent variables, but it is usually limited on the situation when the dependent variable is quantitative variable. Compared with this, that discriminant analysis can be used in the situation when the dependent variable is not quantitative, but it doesn't have the beautiful statistical character of regression analysis, one bridge to link these two methods is Logistic regression.This thesis is beginning with the limited the linear regression on categorical outcome variables, made a complete introduction of logistic regression model through the comparison between logistic regression and normal linear regression. Then get the method to use it in discriminant analysis by analyze the character of logistic regression, it is specially to advise a new method to use logistic regression in discriminant analysis, and we using an example about English learning to examine the efficiency of this method at the end of this thesis.