[問題] 98交大統研統計學
3. (30%)
Let T1 and T2 be sufficient statistics for Θ ,and suppose that T2=g(T1)
for some function g. Let U be an unbiased estimate of Θ and
V1=E(U | T1) , V2=E ( U | T2) . show that MSE(V2)≦MSE(V1) ,where
MSE is the mean squared error.
我覺得這題在我所見考古題裡面，連個尾巴都沒看過，感覺蠻創新的，
所以po出來討論看看，不知道有沒有高手有想法願意分享，感謝。
--
※ 發信站: 批踢踢實業坊(ptt.cc)
◆ From: 140.113.253.57
推
03/16 23:41, , 1F
03/16 23:41, 1F
→
03/16 23:53, , 2F
03/16 23:53, 2F
→
03/16 23:55, , 3F
03/16 23:55, 3F