Re: [R-sig-Geo] spatio-temporal simultaneous autoregression model

2014-08-07 Thread stevenhuyi
Thanks, Corey. That's really helpful. By the way, have u got the R codes from Edzer? :) Yi -- View this message in context: http://r-sig-geo.2731867.n2.nabble.com/spatio-temporal-simultaneous-autoregression-model-tp7586871p7586894.html Sent from the R-sig-geo mailing list archive at Nabble.com

Re: [R-sig-Geo] spatio-temporal simultaneous autoregression model

2014-08-07 Thread stevenhuyi
Thanks, Corey, that's really helpful. By the way, do u get the R codes from Edzer? :) Yes, you can choose a zero inflated poisson or negative binomial likelihood. Also, I had asked about a similar question a while back and Edzer replied with this: http://r-sig-geo.2731867.n2.nabble.com/Questio

Re: [R-sig-Geo] spatio-temporal simultaneous autoregression model

2014-08-06 Thread Corey Sparks
Yes, you can choose a zero inflated poisson or negative binomial likelihood. - Corey Sparks Associate professor Department of Demography The University of Texas at San Antonio 501 West Cesar E Chavez Blvd San Antonio TX 78207 coreysparks.weebly.com 210 458 3166 Latitud

Re: [R-sig-Geo] spatio-temporal simultaneous autoregression model

2014-08-06 Thread stevenhuyi
Thanks Corey. Yeah, what you mentioned is indeed the problem. However, does INLA can deal with the problem of many "zero" values if a count data model is assumed? >You need to be careful with the incidence data if the Gaussian model isn't appropriate, I assume that many >counties have very low p

Re: [R-sig-Geo] spatio-temporal simultaneous autoregression model

2014-08-06 Thread Corey Sparks
You need to be careful with the incidence data if the Gaussian model isn't appropriate, I assume that many counties have very low prevalence, so a count data model may be more appropriate. I would check out INLA, it will do Bayesian models over space and time, there are some tutorials/examples here

Re: [R-sig-Geo] spatio-temporal simultaneous autoregression model

2014-08-06 Thread stevenhuyi
Thanks, Roger. I will try. It's my fault to mistakenly write the spautolm function as a package. what do mean by saying "if your space-time data are balanced". My data is multiple years of prevalence data of schistosomiasis at the county level, combined with data of environmental determinants.

Re: [R-sig-Geo] spatio-temporal simultaneous autoregression model

2014-08-06 Thread Roger Bivand
On Wed, 6 Aug 2014, stevenhuyi wrote: Hi everyone, It seems that there is no specific R package to implement the space-time SAR model. I happend to find an article entitled "Spatio-temporal regression models for deforestation in the Brazilian Amazon" (literature information: STDM 2011, The Inte

[R-sig-Geo] spatio-temporal simultaneous autoregression model

2014-08-06 Thread stevenhuyi
Hi everyone, It seems that there is no specific R package to implement the space-time SAR model. I happend to find an article entitled "Spatio-temporal regression models for deforestation in the Brazilian Amazon" (literature information: STDM 2011, The International Symposium on Spatial-Temporal A