统计与决策2024，Vol.40Issue(10)：52-57,6.DOI:10.13546/j.cnki.tjyjc.2024.10.009
空间权重选择的模拟研究——基于贝叶斯方法
Simulation Study on Selection of Spatial Weight—Based on Bayesian Method
摘要
Abstract
This paper discusses the effect of Bayesian posterior model probability method on the selection of spatial weight matrix under the spatial Durbin model,and then uses Monte Carlo simulation to verify the effectiveness of the method.The results go as the following:The Bayesian posterior model probability method has high accuracy in weight discrimination,and it improves with the increase of sample size and time,and the relationship between the correct discrimination rate and spatial correlation coef-ficient is U-shaped.In the weight selection with strong correlation,the discrimination ability of the method is decreased,but the effect is still better than ML method.In the static spatial weight discrimination of different attributes,the Bayesian posterior model probability can quickly reach 100%.For the selection of dynamic and static spatial weight,the method has a good performance in the case of large and positive spatial correlation.关键词
贝叶斯后验模型概率/蒙特卡洛模拟/空间权重矩阵Key words
Bayesian posterior model probability/Monte Carlo simulation/spatial weight matrix分类
数理科学引用本文复制引用
范真,王西贝..空间权重选择的模拟研究——基于贝叶斯方法[J].统计与决策,2024,40(10):52-57,6.基金项目
国家社会科学基金资助项目(22BJY160) （22BJY160）
国家自然科学基金青年科学基金项目(72303031) （72303031）
