[问答] SVCJ程序求WINBUGS程序 [推广有奖]

61楼

epoh 发表于 2013-4-4 21:09:30

shatian 发表于 2013-4-4 11:55
老师，如果想产生你在49楼给出的图，，，应该怎么做啊？按照你的程序已经运行出来，有迭代图，但是DIC还不 ...

开启log.txt
选取 J[1] ~ J[5038] 存成 svcj_Jmean.txt
选取 Xiy[1] ~ Xiy[5038] 存成 svcj_Xiymean.txt
选取 sig[1] ~ sig[5038] 存成 svcj_sigmean.txt

要显示DIC,将parameters修改为底下:
parameters <- c("mu","theta","k","sigv","muy","rhoj","sigy","muv","rho","lambda")

62楼

huashijiel 发表于 2013-4-11 16:50:17

epoh 发表于 2013-4-4 21:09
开启log.txt
选取 J[1] ~ J[5038] 存成 svcj_Jmean.txt
选取 Xiy[1] ~ Xiy[5038] 存成 svcj_Xiymean.tx ...

老师，您好，我按照您给的svcj程序运行了一下，但是初值的设定有问题，就无法gen inits，希望您解答一下，万分感谢啦

63楼

epoh 发表于 2013-4-11 17:56:28

huashijiel 发表于 2013-4-11 16:50
老师，您好，我按照您给的svcj程序运行了一下，但是初值的设定有问题，就无法gen inits，希望您解答一下， ...

哈哈,你看错了吧,
我没给过svcj程序

64楼

huashijiel 发表于 2013-4-11 18:16:30

epoh 发表于 2013-4-11 17:56
哈哈,你看错了吧,
我没给过svcj程序

哦哦，是之前的某位楼主发出来的，那我贴出来，您看我怎么赋初值呢？我的点gen inits总是出现undefined real result....
model
{
alpha~dnorm(0,1);
beta~dnorm(0,1);
k<-(-beta);
theta<-alpha/k;
mu~dnorm(0,0.04);
isigmav2~dgamma(2.5,0.1);
sigmav2<-1/isigmav2;
muy~dnorm(0,0.01);
isigmay2~dgamma(10,40);
sigmay2<-1/isigmay2;
rho~dunif(-1,1);
rhoj~dnorm(0,2);
imuv~dgamma(10,20);
muv<-1/imuv;
lamda~dbeta(2,40);
v0mean<-(k*theta+lamda/muv);
v0theta2<-(2*lamda-pow(lamda,2)/(pow(muv,2)));
v0~dnorm(v0mean,v0theta2);
vmean[1]<-(k*theta+(1-k)*v0+lamda/muv);
vtheta2[1]<-(sigmav2*v0+(2*lamda-pow(lamda,2))/(pow(muv,2)));
v[1]~dnorm(vmean[1],vtheta2[1]);
muystar<-muy+rhoj*muy;
ymean[1]<-mu+lamda*muystar;
ytheta2[1]<-v0+pow(muystar,2)*lamda*(1-lamda)+sigmay2*lamda;
y[1]~dnorm(ymean[1],ytheta2[1]);
for (i in 1:N){
vmean[i+1]<-(k*theta+(1-k)*v+lamda/muv);
vtheta2[i+1]<-(sigmav2*v+(2*lamda-pow(lamda,2))/(pow(muv,2)));
v[i+1]~dnorm(vmean[i+1],vtheta2[i+1]);}
for (i in 1:N){
ymean[i+1]<-mu+lamda*muystar;
ytheta2[i+1]<-v0+pow(muystar,2)*lamda*(1-lamda)+sigmay2*lamda;
y[i+1]~dnorm(ymean[i+1],ytheta2[i+1]);
}
}

list(y=c(0.678,0.658,0.638,0.618,0.188,0.918,0.338,0.278,0.858,0.348,0.648,0.848,0.708,0.458,0.228,0.648,0.428,0.468,0.738,0.508,0.278,0.948,0.158,0.958,0.138), N=24)

65楼

xuan1993 发表于 2013-7-20 11:06:46

epoh 发表于 2013-4-11 17:56
哈哈,你看错了吧,
我没给过svcj程序

老师您好：可否帮忙给我解决一下问题，谢谢您了，我找了半天都不知道错哪里了？openbugs程序如下:
model
{
alpha ~ dnorm(5,50)
blpha ~ dnorm(5,50)
for( i in 1 : N ) {
p ~ dbeta(alpha,blpha)
}
for( i in 1 : N ) {
r ~ dbin(p,n)
}
feta <- 1 / (1 + alpha + blpha)
}
list(N=47,n=c(1,9,10,6,17,12,9,5,5,2,1,10,1,15,9,12,4,7,8,8,9,2,12,8,4,5,8,9,13,3,2,9,5,3,6,2,3,4,7,5,6,8,6,5,4,4,1),r=c(0,0,0,1,2,1,1,0,1,1,1,2,1,2,1,2,1,2,5,1,4,1,0,1,2,4,2,4,6,1,2,1,1,3,4,2,1,3,3,2,3,5,3,3,3,3,1),p=c(0,0,0,0.167,0.118,0.083,0.111,0,0.2,0.5,1,0.2,1,0.133,0.111,0.167,0.25,0.286,0.625,0.125,0.444,0.5,0,0.125,0.5,0.8,0.25,0.444,0.462,0.333,1,0.111,0.2,1,0.667,1,0.333,0.75,0.429,0.4,0.5,0.625,0.5,0.6,0.75,0.75,1))
list(alpha=1,blpha=1)
前面所有的步骤都OK 但是最后updating时出现了错误：“sorry something went wrong in procedure UpdaterStd.Sample”，也就是通常说的最后无法updating。
本文来自: 人大经济论坛 S-Plus&R专版版，详细出处参考： https://bbs.pinggu.org/forum.php? ... amp;from^^uid=4048590