这很好用:
fit.mc1 <-MCMCglmm(bull~1,random=~school,data=dt1,family="categorical",
prior=list(R=list(V=1, fix=1), G=list(G1=list(V=1, nu=0))), slice=T)
这样做:
fit.glmer <- glmer(bull~(1|school),data=dt1,family=binomial)
但是现在我正在尝试使用包glmmadmb
,但这不起作用:
fit.mc12 <- glmmadmb(bull~1+(1|school), data=dt1, family="binomial",
mcmc=TRUE, mcmc.opts=mcmcControl(mcmc=50000))
它会生成错误:
Error in glmmadmb(bull~ 1 + (1 | school), data = dt1, family = "binomial", :
The function maximizer failed (couldn't find STD file)
In addition: Warning message:
running command '<snip>\cmd.exe <snip>\glmmadmb.exe" -maxfn 500 -maxph 5
-noinit -shess -mcmc 5000 -mcsave 5 -mcmult 1' had status 1
答案 0 :(得分:0)
嗯。任何可重复的例子的可能性......?
以下简单模拟案例似乎有效(尽管glmmADMB
mcmc
慢于MCMCglmm
- 它实际上尚未完成我,虽然似乎没有抱怨但仍然在忙碌着。)
对于这种简单的情况,我怀疑glmmADMB
由MCMCglmm
支配,尽管如果你正在处理反贝叶斯裁判,它可能会有用......
nschool <- 20
nrep <- 20
dt1 <- expand.grid(school=LETTERS[1:nschool],rep=seq(nrep))
set.seed(101)
u.school <- rnorm(nrep)
dt1$eta <- u.school[dt1$school]
dt1$bull <- rbinom(nrow(dt1),size=1,prob=plogis(dt1$eta))
library(MCMCglmm)
fit.mc1 <-MCMCglmm(bull~1,random=~school,data=dt1,family="categorical",
prior=list(R=list(V=1, fix=1), G=list(G1=list(V=1, nu=0))),
slice=TRUE)
library(lme4)
fit.glmer <- glmer(bull~(1|school),data=dt1,family=binomial)
library(glmmADMB)
fit.mc12 <- glmmadmb(bull~1+(1|school), data=dt1, family="binomial",
mcmc=TRUE, mcmc.opts=mcmcControl(mcmc=50000))