作者gone19 (gone19)
看板Statistics
标题[问题] 多元回归的自由度
时间Thu Oct 18 09:58:02 2018
我的笔记是这样～
https://i.imgur.com/9H5B2qa.jpg
我想问
1. 自由度 reduced model 怎麽看？
我的认知中取full df即可
2.我的笔记中
F2,n-5 是怎麽来的
F with (d.f. =k, n-k-1)
Reduced Model: k=X1,X2 =2个
Full Model: k=X1,X2, X3,X4 =4个
So, F k, n-k-1 =F 2,n-5
2=Reduced Model: k=X1,X2 =2个
5=Full Model: k=X1,X, X3,X4 =4个, n-k-1=n-5
Reduced Mode + Full Model 取其可cover 的最大自由度，所以取k=2 for reduced and k=4 for full to cover all range?
这观念中有不正确，请跟我说
感谢?
--
※ 发信站: 批踢踢实业坊(ptt.cc), 来自: 60.250.74.30
※ 文章网址: https://webptt.com/cn.aspx?n=bbs/Statistics/M.1539827885.A.020.html
1F：→ andrew43: 分子df=2就是少了2个系数，分母df就是full的残差df 10/18 15:21
2F：→ andrew43: 和什麽cover没关系，单纯只是比较有几个系数要估 10/18 15:24