标题:
空间误差模型的异常值检验Score Test for Single Outlier in the Spatial Au-to-Regressive Error Model
作者:
戴晓文, 金立斌, 石磊
关键字:
空间误差模型, 异常值检验, 均值漂移模型, 方差加权模型, 得分统计量 Spatial Auto-Regressive Error Model; Outlier Detection; Mean-Shift Outlier Model;
Variance-Weighted Model; Score Test
期刊名称:
《Statistics and Application》, Vol.2 No.4, 2013-12-24
摘要:
本文研究了空间误差模型中单个异常值检验问题。分别在均值漂移模型和方差加权模型这两种异常值模型下给出了得分检验统计量的具体形式及其近似分布。并应用Anselin(1988)的哥伦比亚市犯罪数据进行实例分析，以证实方法的有效性。最后，利用修正模型对异常值进行处理和分析。This paper studies single outlier detection in the spatial auto-regressive error model. Specific formula based on score statistics and their approximate distributions are derived under mean-shift outlier model and variance-weighted model respectively. Analysis of a real example shows that the proposed method is effective for identifying outlier in the spatial auto-regressive error model. Finally a modified model to adjust the outliers is proposed and compared with the original model.