Based On The Glm,Glmm,Pls-logistic,the Current Income Distribution Of Satisfaction In China Is Analyzed

Generalized linear model is a generalization of the linear model,the random error term has been generalized to exponential distribution,it has more flexibility when processing the data,is more extensive.In the actual problem,however,the addition of some variables can lead to a random variable is not independent,it is contrary to the basic hypothesis of generalized linear models.So,this has led to a with a generalized linear model dealing with specific problems may ignore some important variables,which can lead to result in deviation we in dealing with the real problem,the result of the impact analysis.So some scholars to add random effects to the generalized linear model and generalized linear mixed model is established,a move that greatly expand the traditional analysis of the field.In the traditional method of satisfaction analysis policy tend to be some qualitative analysis,this paper based on the research of the traditional analysis method,combining previous research,through the generalized linear model and generalized linear mixed models of quantitative analysis of the national policy on the current satisfaction.In this paper,the data from China people's university of "social mentality and policy assessment associated" research,this topic is based on a questionnaire survey of the nationwide.GLM,GLMM,PLS-logistic introduction policy satisfaction evaluation model,provides a new path for satisfaction evaluation research.In this paper,the main content is as follows:1.In detail in this paper,the design of the questionnaire content as well as the assignment of the various options,based on the research of the questionnaire data,found that is a typical longitudinal data,and previous literature has scholars study satisfaction evaluation of ordinarylogistic regression,has the very high similarity with the generalized linear model.Connection function more diversified,coupled with the generalized linear model is not only to divide the dependent variable into binary discrete variables,so consider a generalized linear model to analyze the degree of satisfaction evaluation problem.In based on GLM analysis,select the part of the representative of the variables in the questionnaire is as follows: the current income distribution in China satisfaction as dependent variable,nationality,residence,level of education,work unit as the independent variable.Dummy variable and the independent variables,and the respondents were divided into groups,60 R software,the application of quantitative analysis for this class of 60 people's satisfaction of the policy situation,has carried on the sort,and analyzes the individual independent variables influence to satisfaction.From the overall view,satisfaction with the current income distribution in China national situation is relatively good,but still expresses the national hope countries in income distribution policy further perfect the incentive constraints mechanism.And analysis found that the education level of high and low,and the degree of satisfaction is directly proportional to the height,so countries should introduce relevant policies to improve the national common level of education.2.The annual household income as a random effects to build satisfaction evaluation is reasonable.Analysis found that from the previous section,through the fixed national,registered permanent residence,level of education that three corresponding argument,research different work the crowd satisfaction,calculation comparison,found that people working in enterprises and retirees higher satisfaction.Associated with the different work units of high and low income,so decided to household income as a random effects added to the model.This section of the independent variable data processing method and the slightly different(see table 3.4),the iterative parameter convergence after 100 times,found the parameters of the whole fitting effect is greatly improved;For a single effect factors as fixed effects of ethnic factors,population factor,level of education factor and work unit factor of parameter estimation is more significant.Especially by the parameters of the degree of education significance a lot of improvement.Also illustrates the GLMM applicability for satisfaction evaluation.3.The partial least-squares regression in the real production with more extensive,mostly used in water quality measurement,medical,biology,geology and other sciences.PLS-logistic model is the ordinary logistic model,the principal component extraction method and typical correlation analysis to combine a new analysis model,cleverly put the advantages of partial least-squares regression and logistic regression.PLS-logistic model can eliminatethe multicollinearity problem between the observed data,at the same time in the extract when considering the relationship between independent variable and dependent variable.Inspired by our predecessors,this article will first PLS-logistic model used for analysis of empirical satisfaction.Table 3.7 in this paper in detail the selection of variables,using statistical software of matlab model building,the iteration can be found in order to extract the constituent of three partial least squares model when the most to achieve optimal,finally,in order to further analyze the independent variable to explain the satisfaction function,calculated the VIP importance index to measure,analysis and gives the policy Suggestions.