空间误差模型的多个异常值得分检验

Score Test for Outliers in the Spatial Autoregressive Error Model

本文研究了空间误差模型(SEM)中多个异常值的检验问题,基于均值漂移模型和方差加权模型这两种异常值模型给出了得分检验统计量的具体形式及其渐近分布。并应用实例分析验证了检验统计量的有效性,最后给出了修正模型的方法。

This paper studies the score test for multiple 0utliers in the spatial autoregressive error models （SEM）. The formulae and approximate distribution of score statistics are obtained under both mean-shift outlier model and variance-weighted model. A real example is analyzed and the result shows that the proposed methodology is useful for identifying multiple outliers in the spatial autoregressive error models. Finally, outlier-correction models are proposed to accommodate outliers in the data set.