[問題] 有關MLE

[LITTLEN (有沒有那麼雖阿~~~)]

這個MLE問題有點難, 只想問看看有沒有人有想法, 謝謝..

Let (Yi1, Yi2), i=1...n be a sequence of indpendent pairs of independent

Bernoulli variables with

        logit P(Yi1=1) = αi
        logit P(Yi2=1) = αi+β

for unspecified parameters β and αi, i=1...n

(i)compute, approximately the asymptotic bias in the MLE for β, as n --> inf

(ii)Suggest an approach to asymptotically unbiased estimation of β

logit P(Yi1=1) = αi 表示 P(Yi1=1) = e^(αi)/(1+e^(αi))

--
※ 發信站: 批踢踢實業坊(ptt.cc) 
◆ From: 160.39.5.80
→ bmka:你的想法呢?