中文摘要：

文章在对参数回归及非参数回归方法进行介绍的基础上,采用线性回归模型、引入虚拟变量的非线性回归模型以及N-W核权函数回归估计,对1983—2014年我国外商直接投资总额与GDP关系进行探讨。最后分别给出3类方法的拟合值以及MAE和MSE。从拟合效果来看,相对于线性回归模型,引入虚拟变量的非线性回归模型以及非参数回归方法更能有效地拟合FDI与GDP之间的关系,且窗宽为0.25的非参数估计效果更优。

英文摘要：

On the basis of the introduction of parametric regression and nonparametric regression,the linear regression model,the nonlinear regression model with dummy variables,and N-W kernel weight function regression were used to study the relationship between FDI and GDP from 1983 to 2014.Finally,the fitting values of the three methods and MAE and MSE are given.From the fitting effect,as opposed to the linear regression model,nonlinear regression model with dummy variables and nonparametric regression method are more effective in fitting the relationship the relationship between FDI and GDP,and the window width is 0.25 is better.