Analysis Of Several Economic Data By Using Nonparametric Methods

One primary application of Statistics is to seek an appropriate model based on observational data. There are many ways to fit data, of which linear regression model and generalized linear models might be the more classic model. In recent years, due to the complexity of variables and the developing of computer, it is extended to non-linear model. If certain variables based on experience to know there is some functional relation can be used, we can find a function of some parameters to fit the observed data, which is called parametric model. But when we have the intrinsic relationship between variables is not clear, the use of parametric model will bring a larger model error, so nonparametric and semiparametric models are introduced. Compared with the parameter model, nonparametrica and semi-parametric model of the functional relationship between variables less restrictive, and adaptable, more realistic problems, it has also become a recent hot topic to study one or two decades.As to nonparametric and semiparametric regression models, the most widely used regression methods include: local smoothness, including Nadaraya-Watson kernel estimate, Gasser-Muller estimate, local polynomial method; spline method(such as smoothing spline, spline penalies and so on); other is orthogonal series estimates, such as Fourier series and so on.This article discussed mainly Nadaraya-Watson kernel estimation method for nonparametric model estimation, estimate coefficient functions from varying coefficient model by using local polynomial method, including estimate method. To illustrate the advantages of estimation methods, we will use nonparametric model to analysis the functional relationship between consumer spending and disposable income, fit and Forecast the data of the per capita GDP in Xinjiang, which includes some graphics such as histogram, box plot and estimated function curve.