作者joanne4816 (Joanne)
看板Statistics
标题[问题] Bayes procedure和Frequentist test问题
时间Sun May 25 02:32:31 2014
想要请教大家下面这道题目:
Use the binomial model for X with n=25; 0<θ<1, A={a1,a2} where a1 means:“
say θ<=0.2”and a2 means:”say θ>0.2”.
a)Let the loss be L(a1,θ)= 0, ifθ<=0.2 or 5, ifθ>0.2
L(a2,θ)= 10, ifθ<=0.2 or 0, ifθ>0.2
Find the Bayes procedure for a Beta prior with α=6 and β=18
b)Consider the following class of decision procedures:
t(x)=a2 if and only if X>C where C can be any value in {0,1,2,…,n-1}
Find C such that the defined procedure is a frequentist test of size α=0.05
c)Show that there is a Beta prior such that the corresponding Bayes procedure
is the frequentist test in (b).
d)Assume the prior is Uniform on the interval 0 to 1. Show that there is a
very drastic modification of the loss function above which makes the
frequentist test in (b) a Bayes procedure wrt this prior.
下面是我现在遇到困难的地方~
a)我有解出到 Bayes Procedure t(x)=a1 iff 1/3<=P(θ<=0.2|x)
a2 otherwise
不知道要如何把x解出来
b)从这条式子P(X>C|θ=0.2)<=0.05有解出C=8
不知道对不对
c)是指说改变Beta(α,β)中的α和β让Bayes procedure 变成 frequentist test吗?
d)就完全没有头绪了
可以请大家给点想法吗?
谢谢大家^^
--
※ 发信站: 批踢踢实业坊(ptt.cc), 来自: 128.211.253.136
※ 文章网址: http://webptt.com/cn.aspx?n=bbs/Statistics/M.1400956354.A.FC6.html