从贝叶斯随机效应模型（JAGS）中提取预测

Question

我是贝叶斯人的新手，我正试图提取预测结果和可靠的区间以进行绘图，但我还不太清楚如何编码。理想情况下，我既希望总体预测（例如两次治疗之间比较dist_feed的预测值），又要明确包含其他变量（例如在每次治疗中比较雄性和雌性以及在每次治疗中雌性与雌性嵌套）的预测值）。任何提示将不胜感激。

型号：

sink("dist_feeder.jags") 
cat("
model {

for( r in 1 : Ndata ) {
dist_feeder[r] ~ dnorm(mu[r] , tau) 
mu[r] <-  alphanobo[nobo[r]] + b0[nobo[r]]*treatment[r] + b1[nobo[r]]*sex[r] + b2[nobo[r]]*nest[r] +b3[nobo[r]]*nest[r]*sex[r] + b4[nobo[r]]*brood[r]+ b5[nobo[r]]*brood[r]*sex[r]
}


for (i in 1:Nnobo) { alphanobo[i] ~ dnorm(0, taunobo)
                      b0[i] ~ dnorm(mutreatment,tautreatment)
                      b1[i] ~dnorm(mu1G , tau1G )
                      b2[i] ~dnorm(mu2G , tau2G)
                      b3[i] ~dnorm(mu3G , tau3G )
                      b4[i] ~dnorm(mu4G , tau4G )
                      b5[i] ~dnorm(mu5G , tau5G )
                      }

munobo ~ dnorm(0,0.01)

tau <- 1/(sigma*sigma)
sigma ~ dunif(0,500) 

taunobo <- 1/(sigma.nobo*sigma.nobo)

sigma.nobo ~ dunif(0,500)

mutreatment ~ dnorm(0,0.01) 
tautreatment ~ dgamma(0.1,0.1)

mu1G ~ dnorm(0,.01)
tau1G ~ dgamma(.1,.1)

mu2G ~ dnorm(0,.01)
tau2G ~ dgamma(.1,.1)

mu3G ~ dnorm(0,.01)
tau3G ~ dgamma(.1,.1)

mu4G ~ dnorm(0,.01)
tau4G ~ dgamma(.1,.1)

mu5G ~ dnorm(0,.01)
tau5G ~ dgamma(.1,.1)

 }

",fill = TRUE)
sink()


 ni <- 55000
 nt <- 1
 nb <- 25000
 nc <- 3
 parameters <- 

 c("munobo","sigma", "alphanobo","sigma.nobo","mutreatment", 
   "tautreatment","b0","b1", "b2","b3","b4","b5","mu1G","mu2G", 
   "mu3G","mu4G","mu5G","tau1G","tau2G","tau3G","tau4G","tau5G")

mod1 <- jagsUI(jags.data, inits=NULL, parameters, "dist_feeder.jags", n.chains = nc, n.thin = nt, n.iter = ni, n.burnin = nb)

尝试使用预测提取代码：

pred_fdist <-matrix(nrow=80000,ncol=2)

for(i in 1:80000){
 pred_fdist[i,] <- mod1$sims.list$mutreatment[i] + mod1$sims.list$munobo[i] + mod1$sims.list$mu1G[i] + mod1$sims.list$mu2G[i] + mod1$sims.list$mu3G[i] + 
mod1$sims.list$mu4G[i] + mod1$sims.list$mu5G[i]
}