Combination Of Nonparametric Regression Model Method And Application

Objective understanding of the phenomenon behind the nature and described thelaw has always been a statistically’s development objectives, statistical regressionmodel is the tool of a good understanding and elaboration of the law. However, in thestatistical model analysis of economic phenomena and other social problems, oftenencountered in the choice of parametric regression models or nonparametricregression model, and decided to choose a non-parametric regression model, willappear in a variety of non-parametric regression model of how to choose a bettermodel. This thesis focuses on the two issues, the use of instances of attempts to modelthe pros and cons of the use of a variety of non-parametric regression methods tomodel examples to reflect the specific issues. On this basis, the paper presents anonparametric regression model combines the benefits of a combination ofnon-parametric regression model law, give full play to the role of the robustness of thenonparametric regression analysis of the instance was found that the combination ofnon-parametric regression method better reflect the actual situation, and achieved thedesired results.Paper on how to choose the regression model parameters and nonparametricregression model reference opinions are put forward. The parameters of the traditionalregression model judge the model’s good or bad with minimum error squares.Research found that for nonparametric regression model speaking, can’t again withminimum error square for evaluation standard, for a set up economic model, firstlyevaluation standard is model conforms to economic theory significance. In theeconomic theory analysis, often meet this question: The independent variablecoefficient model with different phase change, the parameters of regression model inthe treatment of such problems represented very weak, often can only give a singleconstant coefficient, cannot reflect the nature of the problem, at this time,nonparametric regression model advantage to reflect, should be first selectionnonparametric regression model. Secondly, economic theory analysis also oftenappear such problems, namely between the two variables in theory should haveobvious causal relationship, and the parameters of the regression model is establishedshows its coefficient was not through the significant inspection, where you should select the nonparameter model to fitting data, to the variable relationship between thegood explanation.