摘要
本文对Dorfman群检验问题建立一个Bayes模型,取NC项比率ω的先验分布为Beta分布。利用共轭分布法,我们得到的后验分布还是Beta分布。然后由Bayes估计得出ω由此通过群检验本身过程就可得出ω。而且可以随着检验次数的增加而逐步调整。节省了检验次数,同时又没损失信息。本文还给出了一种利用专家意见确定Beta分布参数的方法讨论了最佳分组的步骤与方法。
Several problems on the statistical effects of group testing errors are discussed and some statistical consequences of group testing are described. A Bayesian model for the effects of group testing errors is presented. It is assumed that the ratio of Nc items in a lot is a random variable with a Beta distribution. Optimal sequence of group for a given lot size,using a beta prior distribution for the ratio of Nc items in a lot is derived.
