非正态分布下具有自回归误差项的空间自回归模型变量选择研究

将变量选择引入空间计量模型,讨论具有自回归误差项的空间自回归模型的变量选择问题。在残差非正态独立同分布的条件下,通过最大化信息熵,提出空间信息准则,并证明其在该模型变量选择中具有一致性。模拟研究结果表明:无论对单个系数还是对全部系数,空间信息准则都能很好识别,且与经典的赤池准则相比具有较大的优势。因此,空间信息准则是一种更为有效的变量选择方法。

Research on Variable Selection in Spatial Autoregressive Model with Autoregressive and Non-Normal Disturbances

Introducing variable selection in spatial model ,we consider how to select variable in Spatial Autoregressive Model with Autoregressive Disturbances .Based on the assumption that the residuals are independently and identically distributed ,we obtain Spatial Information Criterion (SIC) by minimizing information entropy .We prove the selection consistency of the introduced criteria and evaluate their performance by Monte Carlo simulation .The results suggest that no matter for one or for all coefficients , they can be recognized by SIC .What's more ,SIC is better than AIC for the variable selection of spatial model .Therefore ,SIC is a much efficient one .