作者tokyo291 (工口工口)
看板Statistics
标题[程式] SAS proc genmod的警告
时间Sat Oct 4 18:48:53 2014
[软体程式类别]:
SAS
[程式问题]:
由於为sparse data 产生以下警告讯息
WARNING: The generalized Hessian matrix is not positive definite. Iteration
will be terminated
[软体熟悉度]:
高(1年以上)
[问题叙述]:
再跑GEE模型分析的时候,由於资料在某个时间为sparse data
SAS无法估计,有试过加改maxiter修改收敛次数,可是依然出现警告讯息
想请问是否有其他的option可以调整,以解决此问题
[程式范例]:
PROC GENMOD data=patient2 DESCEND ;
CLASS group (ref="2") sex SN PreVac40_B(ref="0");
MODEL SR_all= Age Kt_V group PreVac40_B BUN Ferritin time /Dist=bin
link=logit maxiter=100000 ;
REPEATED SUBJECT=SN/type=exch;
run;
--
※ 发信站: 批踢踢实业坊(ptt.cc), 来自: 140.116.52.59
※ 文章网址: http://webptt.com/cn.aspx?n=bbs/Statistics/M.1412419735.A.5EA.html
1F:→ bmka: outcomes in some subgroups are all zero or one. 10/04 20:30
2F:→ bmka: or the subgoups defined by a covariate are subsets of 10/04 20:35
3F:→ bmka: those defined by another covariate. You might want to 10/04 20:35
4F:→ bmka: pay special attention to PreVac40_B (I suppose this 10/04 20:36
5F:→ bmka: indicates that a subject's baseline antibody titer is 10/04 20:37
6F:→ bmka: >=1:40), and other covariates that may affect antibody 10/04 20:38
7F:→ bmka: level, including gender and age. 10/04 20:38
感谢您的回覆,您提的这两个问题我们都有想过,
现阶段想尝试解决第一个可能的问题-sparse data
不过查过许多SAS mannual和其他人的code都找不到解决方案...
※ 编辑: tokyo291 (140.116.52.59), 10/05/2014 19:55:01
8F:→ bmka: 如果大部份的observation都是0,为什麽硬是要fit这个model 10/05 22:42
9F:→ bmka: Consider reduce the number of covariates. 10/05 22:43