Statistical Prediction Based On Generalized Linear Model And Its Application

Regression analysis is a widely used quantity analysis method. When it is used to analyze the relation between dependent variable and independent variables, the basic condition is that the dependent variable should be continuous numerical variable. However, it won't be always like that in our practical model. When the dependent variable is not continuous numerical variable, the ordinary regression model can't be used any more.Generalized regression model is just one of the statistical methods that can solve this problem. It fits for both continuous and discrete data, especially for the discrete data. This paper makes a study on the generalized linear regression model with its model equation, the method of parameter estimation, significance test of the parameter and model fitting test. Meanwhile, this paper builds a generalized model on the Olympus after service satisfaction, and analyzes for the model, which indicates the superiority of generalized model when the variables are not continuous numerical variables.