[程式] sas跑gee(廣義估計方程式)

看板Statistics作者 (take it easy)時間14年前 (2010/07/15 16:26), 編輯推噓3(3推 0噓 13→)
留言16則, 4人參與, 最新討論串1/1
[軟體程式類別]:sas [程式問題]:gee 迴歸 [軟體熟悉度]:低(1~3個月) [問題敘述]:輸入指令如程式範例範例所示 輸出的結果中,我看不到r-square 及 vif 不知道模型的解釋力如何&是否存在共線性問題 請問我是不是忽略了什麼指令呢? 謝謝^^ [程式範例]: proc genmod data=try; class id /desc; model count_1=id_g1-id_g3 period1 id_g1*period1 id_g2*period1 id_g3*period1 id_sex1 id_age_g1-id_age_g3 id_home1-idh_home3/d=normal; repeated subject=id/ type=UN; run; ----------------------------------------------------------------------------- -- ※ 發信站: 批踢踢實業坊(ptt.cc) ◆ From: 140.112.118.38
推
07/15 21:31, , 1F
找vif不需要用到genmod, proc reg就可以了
07/15 21:31, 1F
→
07/15 21:48, , 2F
我後來有用proc reg去試跑~只是想知道在gee的報表裡可否
07/15 21:48, 2F
→
07/15 21:49, , 3F
呈現....而且沒有r-square讓我很苦惱@@"請大家幫忙囉~
07/15 21:49, 3F
推
07/15 21:59, , 4F
都已經是genmod了，看的都是一些比較進階的東西例如deviance
07/15 21:59, 4F
→
07/15 22:00, , 5F
真的非r-sq不可的話就請去找其他regression procedure比較快
07/15 22:00, 5F
→
07/16 08:05, , 6F
好的~謝謝指教!所以genmod的語法沒有計算r-square是正常吧
07/16 08:05, 6F
推
07/16 15:17, , 7F
可以看working correlation
07/16 15:17, 7F
→
07/20 09:51, , 8F
不知是否我記錯working correlation 是指重複測量觀察值
07/20 09:51, 8F
→
07/20 09:52, , 9F
之間的關係?
07/20 09:52, 9F
→
07/20 10:03, , 10F
求係數時 所用的working correlation matrix
07/20 10:03, 10F
→
07/20 10:07, , 11F
The method relies on the independence across subjects
07/20 10:07, 11F
→
07/20 10:08, , 12F
to consistently estimate the variance of the proposed
07/20 10:08, 12F
→
07/20 10:08, , 13F
estimators even when the assumed working correlation
07/20 10:08, 13F
→
07/20 10:09, , 14F
structure is incorrect. Zeger (1988), Zeger, Liang,
07/20 10:09, 14F
→
07/20 10:09, , 15F
and Albert (1988), and Liang, Zeger, and Qaqish (1992)
07/20 10:09, 15F
→
07/20 10:09, , 16F
provide further detail on the GEE methodology. (轉貼)
07/20 10:09, 16F
‣ 更多 minna05 的文章
文章代碼(AID): #1CFiOdJX (Statistics)