Hi All,
I'm trying to use the geoRglm package to run a poisson spatial glm
on a dataset with several covariates. When I run without covariates I
have no problems.
control1.data.geo <- mcmc.control(S.scale=0.2, thin = 1)
model1.data.geo <- list(cov.pars = c(1,1), beta=c(1),
family="poisson")
test1.model1 <- glsm.mcmc(data.geo, model=model1.data.geo,
mcmc.input=control1.data.geo)
iter. numb. 1000 : Acc.-rate = 0.634
MCMC performed: n.iter. = 1000 ; thinning = 1 ; burn.in = 0
When I try to run model tuning with more than one variable specified
in the model using trend spatial I get an issue with accept-rates
always equal to 0. Am I parameterizing the model incorrectly?
control1.data.geo <- mcmc.control(S.scale=0.2, thin = 1)
trend.max.trans <- trend.spatial(trend= ~ Maxent + TransLen, data.geo)
model1.data.geo <- list(cov.pars = c(1,1), beta=c(1,1,1),
family="poisson", trend=trend.max.trans)
test1.model1 <- glsm.mcmc(data.geo, model=model1.data.geo,
mcmc.input=control1.data.geo)
prep1.fit1.model1.data.geo <- prepare.likfit.glsm(test1.model1)
iter. numb. 1000 : Acc.-rate = 0
MCMC performed: n.iter. = 1000 ; thinning = 1 ; burn.in = 0
Thanks
Ken
[[alternative HTML version deleted]]
