Questions tagged «rao-blackwell»

2
为什么Rao-Blackwell定理要求?
Rao-Blackwell定理指出 让是一个估计与所有。假设对于是足够的,并且让然后对于所有,不等式是严格的,除非是的函数θ^θ^\hat{\theta}θθ\thetaE(θ^2)&lt;∞E(θ^2)&lt;∞\Bbb E (\hat{\theta}^2) < \inftyθθ\thetaTTTθθ\thetaθ∗=E(θ^|T)θ∗=E(θ^|T)\theta ^ * = \Bbb E (\hat{\theta}|T)θθ\thetaE(θ∗−θ)2≤E(θ^−θ)2E(θ∗−θ)2≤E(θ^−θ)2\Bbb E (\theta^* - \theta )^2 \leq \Bbb E (\hat{\theta} - \theta )^2θ^θ^\hat{\theta}TTT 如果我正确理解了该定理,则表明如果我有足够的统计量来计算,则给定的的条件期望值就是(\ hat {\ theta}-\ theta)^ 2TTTθθ\thetaθ^θ^\hat{\theta}TTTminθ^Eminθ^E\min_{\hat{\theta}} \Bbb E (θ^−θ)2(θ^−θ)2(\hat{\theta}-\theta)^2 我的问题 我是否纠正θ∗θ∗\theta^*最小化EE\Bbb E (θ^−θ)2(θ^−θ)2(\hat{\theta}-\theta)^2? 为什么Rao-Blackwell定理要求E(θ^2)&lt;∞E(θ^2)&lt;∞\Bbb E(\hat{\theta}^2) < \infty? 为什么不等式严格,除非θ^θ^\hat{\theta}是T的函数TTT?
By using our site, you acknowledge that you have read and understand our Cookie Policy and Privacy Policy.
Licensed under cc by-sa 3.0 with attribution required.