摘要
在一个参数的可选先验分布类中选择一个最优先验密度的问题,类似于从参数空间中估计一个恰当参数的问题,或类比为从决策类中选择出最优决策的问题.从这一角度出发,就可以借助统计学中已有方法,建立一套关于先验分布选取的合理方法,同时能估计其精度.
Selecting an optimal prior density problem in a series of selectable prior distributions of a parameter is similar to estimating a suitable parameter from parameter space, or selecting a optimal decision in a series of decision problems. From this point, we can establish a set of reasonable methods about selecting prior distribution and estimate its accuracy with the help of existed methods in statistics.
关键词
先验分布
优选
后验误差
后验期望损失
prior distribution
optimal election
posterior error
posterior expected loss
