作者yuu0826 (要来念请慎重考虑)
看板Statistics
标题[计算] signal-detection的计算
时间Thu Apr 26 14:13:25 2007
其实我不知道该去哪里问...因为是实验法的题目
好像又不属於统计的范畴@@" 而且看到指数对数就很头痛 >//////<
所以想请教大家 给我一些指引 :)
Suppose a gang of smugglers were trying illegally to export radioactive
material.In an effort to try to stop this, you monitor with a Geiger
counter every package leaving the country.
For those packages without the radioactive material (NRA)
the number of counts is given approximately by an exponential
distribution.
Letting x equal the number of counts:
f(x|NRA) = e^(-x)
For the package with radio active material (RA)
f(x|RA) = λe^(-λx) 0 < λ < 1.00
(a) What is the likelihood ratio of a package’s containing radioactive
material as opposed to its not containing radioactive material?
(b) If the a priori probability of a package’s containing radioactive
material is 1 in 10^6 , and if
$100 is cost of opening a package and repacking it if does not contain
radioactive material (false alarm -$100)
$10^4 is value of catching a radioactive package (Hit +$10^4)
$10^6 is cost of letting a radioactive package through (Miss -$10^6)
$0 is the value of letting a nonradioactive package through
(correct rejection +$0)
What value of likelihood ratio should lead one to ask that the package
be opened?
(c) If =1/6, what value of x would one need to attain that value of
likelihood ratio in (b)?
Suppose your budge director gives you $300,000 per year, so that you can
afford to open 3000 packages per year even if none is radioactive.
(d) What number of counts would you use on the assumption that 10
packages are processed each year?
(e) What is your probability of detecting a radioactive package if=1/8?
我很努力的想 也只可能知道第一提可能的解答是什麽
我会再努力看看 也希望板上的大家可以给我一点指引..
真的很感谢你们^^
--
--
※ 发信站: 批踢踢实业坊(ptt.cc)
◆ From: 125.232.114.148
--
--
※ 发信站: 批踢踢实业坊(ptt.cc)
◆ From: 125.232.114.148
※ 编辑: yuu0826 来自: 125.232.114.148 (04/26 14:13)
※ 编辑: yuu0826 来自: 125.232.112.5 (04/27 11:30)