广义空间异常值修正模型及应用

General Spatial Outlier Modified Model and Its Applications

本文针对广义空间模型中出现的异常值提出了相应的修正模型。分别基于广义空间模型的均值扰动和方差扰动下提出了对应的异常值修正模型，并给出了参数估计方法。通过将该方法应用于中国能源利用效率的区域特征问题，分析了其中的异常值并建立了异常值修正模型。该修正模型能有效改进模型拟合效果，为处理数据中出现的异常值提供了一种新的思路。

This paper suggest the outlier modified models when there exist outliers in general spatial models. Based on mean-shift and variance-shift perturbation scheme in general spatial models, we pro- posed outliers modified models respectively. The estimating methods of parameters in outliers modified models are given. The proposed method are applied to the study of the regional characters of energy effectiveness in China, and we find that the outlier modified model can effectively improve the model fitting. This method provides another way to deal with outliers in the data.