I am doing on Spatio-temporal diseases mapping.
I want to use inseparable space-time model with four types of interactions which has been proposed by Knorr-Held , 2000.
I am using R2WinBUGS/WinBUGS, but I was wondering how to code the prior distribution for the interaction term taking into account the Kronecker product between the two main effects structure matrix. The interaction between the spatial and temporal effects is posted below. Is there any one who have any hint to about my question.
Hi,
Although it is perfectly possible to implement such a large models in BUGS.
It will require a lot of work. I would suggest you to have a look at the INLA project http://www.r-inla.org/. It provides a more efficient way to fit space-time models using the Bayesian paradigm. You can find some code example of the Held's approach in this webpage http://www.leg.ufpr.br/doku.php/publications:papercompanions:wbonat. It is in Portuguese, but you still can use the R code.
Hope it helps.
Best wishes.
Wagner
Dear Wagner,
Thank you very much for sharing that you have. It is really helpful.
I want to make a comparison between MCMC and INLA results.
With best regards,
Worku
- Badges
- Science topic