混合地理加权回归模型作为一类能简单有效解决空间非平稳问题的数据分析方法已经得到了广泛的应用。在利用该模型分析实际数据时,一个或多个特殊观测点的存在能导致估计结果发生较大改变。为了能有效检测出异常点,系统研究这类半参数模型的统计诊断与影响分析。首先基于数据删除模型定义了参数分量对应的Cook统计量,其次,基于均值漂移模型讨论了异常点的检验问题,构造了相应的检验统计量。
Mixed geographically weighted regression model is frequently used in spatial analysis. In this paper, diagnostic and influential analysis on this semi - parametric model is concerned. A type of Cook' s distance of parametric component based on data deletion model was introduced, and a test statistic for testing whether a given observation is an outlier to the model based on the mean shift outlier model was constructed.