Dear Bugsers
I'm trying to implement a CAR model within a Poisson model in order to explain the extra-poisson variability of relative risks within k small areas.
The model I use is:model {
for(i in 1:N){
Y[i] ~ dpois(theta[i])
theta[i] <- exp(psi[i])*E[i]
psi[i] <- beta0 + alpha*X[i] + b[i]
relativerisk[i] <- exp(psi[i])
}
alpha ~ dnorm(0, 10000)
b [1:N] ~ car.normal(adj[] , weights[] , num[] , tau)
beta0 ~ dflat()
for(j in 1:sumNumNneigh){weights[j] <- 1}
tau~dgamma(0.5 , 0.0005)
}
list(E=c(2.9 , 7.4 , 7.4 , 11 , 7.1 , 6.1 , 3.9 , 10 , 7.5 , 7.2 , 6.4 , 5.6 , 3 , 6.2 , 5.3 , 10.1 , 10.2 , 7 , 4 , 5.6 , 2.1 , 6 , 6.6 , 5.2 , 2.5 , 2.5),
Y=c(1 , 2 , 2 , 4 , 9 , 2 , 2 , 5 , 3 , 3 , 4 , 2 , 0 , 1 , 3 , 3 , 10 , 5 , 3 , 3 , 2 , 2 , 4 , 3 , 1 , 0) , X=c(388.35,491.24,586.22,634.61,860.23,981.95,902.56,758.9,592.44,385.2,254.33,236.83,483.53,187.70,147.35,320.08,529.44,502.35,738.72,887.47,729.92,588.29,501.05,419.83,737.28,813.66),
num=c(3,5,3,4,3,3,5,8,5,8,6,5,2,6,6,7,6,6,7,4,4,3,5,6,3,3) ,
adj=c(
2,12,13,
1,3,10,11,12,
2,4,10,
3,8,9,10,
6,7,8,
5,7,8,
5,6,8,19,20,
4,5,6,7,9,17,19,20,
4,8,10,17,19,
2,3,4,9,11,15,16,17,
2,10,12,14,15,16,
1,2,11,14,15,
1,14,
11,12,13,15,24,25,
10,11,12,14,16,24,
10,11,15,17,18,23,24,
8,9,10,16,18,19,
16,17,19,21,22,23,
7,8,9,17,18,20,21,
7,8,19,21,
18,19,20,22,
18,21,23,
16,18,22,24,26,
14,15,16,23,25,26,
14,24,26,
23,24,25),
sumNumNneigh=126,
N=26)
list( alpha=0, tau=0.5, b=c(0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5),beta0= 0.01)
The problem is: when I run the model I see a window with this message: undefined real result.
I know that is an overflow problem but I changed initial values for nodes.....without any result.
Is there someone who could suggest me better initial values?
Thanks a lot.
Pietro
--
TiscaliNet, libero accesso ad Internet.
http://www.tiscalinet.it
-------------------------------------------------------------------
To mail the BUGS list, mail to [log in to unmask]
You can search old messages at www.jiscmail.ac.uk/lists/bugs.html
To leave the BUGS list, send LEAVE BUGS to [log in to unmask]
If this fails, mail [log in to unmask], NOT the whole list
This list is for discussion of modelling issues and the BUGS
software. For help with crashes and error messages, first mail
[log in to unmask]
|