部分线性空间自回归模型的似然比检验

Likelihood Ratio Test of Partial Linear Spatial Autoregressive Model

考虑部分线性空间自回归模型的检验问题,结合B样条逼近方法和矩阵的正交投影技术,对模型的参数分量的检验问题提出了一种基于正交投影的广义似然比检验方法。并且进一步对检验统计量P值的计算问题,提出了一种基于Bootstrap自助法的迭代算法。在对模型参数分量的检验过程中利用了基于矩阵QR分解的正交投影技术,保证了对参数分量的检验过程不受非参数分量的影响,因此具有较好的适应性。

Considering the test of partial linear spatial autoregressive model,combined with B-spline approximation method and orthogonal projection technology of matrix,a generalized likelihood ratio test method based on orthogonal projection is proposed for the test of parameter components of the model.Furthermore,an iterative algorithm based on bootstrap method is put forward to calculate the P value of test statistics.In the process of testing the parameter components of the model,the orthogonal projection technology based on matrix QR decomposition is used to ensure that the testing process of the parameter components is not affected by the nonparametric components,so it has good adaptability.