Re: [課業] MLE?
※ 引述《ljta (ljta)》之銘言:
: ※ 引述《jasonkeen (Nothing but Net)》之銘言:
: : T is sufficient statistic
: : then f(x1,...,xn;θ) = g(t;θ)h(x1,...,xn|t)
: : where t = T(x1,...,xn)
: : hence, finding θ to maximize f(x1,...,xn;θ)
: : => finding θ to maximize g(t;θ)
: : ︿
: : assume that θ maximize g(t;θ)
: : (g(t;θ) is function of T, and also be function of θ)
: : ︿
: : if MLE is unique, θ = MLE is function of T
: : ︿
: : if MLE is not unique, θ is one of MLE
: : we can't say MLE must be function of T
: http://0rz.tw/36539
: http://0rz.tw/e152b
: 想找論文請透過學校ip或至圖書館登錄
第二個連結好像有人不能開?
Maximum Likelihood and Sufficient Statistics
D. S. Moore
The American Mathematical Monthly, Vol. 78, No. 1 (Jan., 1971), pp.50-52
Published by: Mathematical Association of America
--
※ 發信站: 批踢踢實業坊(ptt.cc)
◆ From: 140.114.36.66
討論串 (同標題文章)