GSLM in binary data set
Ok Paulo, i really appreciate your time and your answer was very helpful!!! Carolina Lang Abarz??a Investigador Instituto de Fomento Pesquero Av. Blanco #839, Valparaiso, Chile. Fono: (32) 2151572 www.ifop.cl ----- Mensaje original ----- De: "Paulo Justiniano Ribeiro Jr" <paulojus at c3sl.ufpr.br> Para: "carolina lang" <carolina.lang at ifop.cl> CC: r-sig-geo at r-project.org Enviados: Mi??rcoles, 5 de Febrero 2014 9:03:40 Asunto: Re: [R-sig-Geo] GSLM in binary data set Carolina The MCMC engine behind geoRglm needs to be tunned. S.scale is the tunning parameter for the spatial random effects. It need to be choosen such that acceptence rates (reported by the following functions to be used) are about 60-65%. So don't expect to get it right at a first try, this is a try and error exercise The other parameters are initial guesses and it would be better of compatible with your data. For instance you could use values from a standard GLM to set beta, the overdispersion to set sigmasq and a variogram of the residuals for phi. The latter needs to be compatible with the distances within the study area so alternativelly you could try something 1/10 of the maximum interpoints distances
On Mon, 3 Feb 2014, carolina.lang at ifop.cl wrote:
Hello everybody, I??m new in this forus and first i will like to congratulate for this space where people share opinions and knowledge. Well, i??m working with a binary spatial data set that were collected in an hydroacustic survey, and i want to apply GLSM, but i have some queries i took this example routine: 1#require(geoR);require(geoRglm) 2#van10<-as.geodata(area11,coords.col=8:9,data.col=10) 3#model.5 = list(cov.pars=c(1,1),cov.model='exponential',beta=1,family="binomial") 4#mcmc.5 = mcmc.control(S.scale = 0.25, n.iter = 30000, burn.in=50000, thin = 100) How do i define the parameters in thirth line (cov.pars and beta) and fourth line ( s.scale, burn.in and thin). I suppose that change in this parameters affects final results. Anyway i don??t have any graphics about spatial correlation in data and estimate a priori parameters like in classical geostatistics (experimental variogram). greetings, Carolina ------------------------------------------------------------------------------------------------ Certificaci?n ISO 9001/2008: Sistema de Datos Bio-Pesqueros ( Arica, Iquique, Coquimbo, Valpara?so, San Antonio, Talcahuano y Calbuco, pesquer?as industriales y artesanales) [[alternative HTML version deleted]]
------------------------------------------------------------------------------------------------ Certificaci?n ISO 9001/2008: Sistema de Datos Bio-Pesqueros ( Arica, Iquique, Coquimbo, Valpara?so, San Antonio, Talcahuano y Calbuco, pesquer?as industriales y artesanales)