Saturday, December 04, 2010

Comparison of results

I am doing a simple comparison of different estimation procedures in dealing with a simple binomial model. Here is where I got started:
---------------------------------------------

library(INLA)
library(npmlreg)
library(MCMCglmm)
library(DPpackage)


data(Seeds)


# Using INLA
formula = r ~ x1*x2 + f(plate, model="iid")
mod.inla = inla(formula, data=Seeds, family="binomial", Ntrials=n)
summary(mod.seeds)


# Using npmlreg
mod.ml <- alldist(cbind(r, n-r) ~ x1*x2 , random=~1, data=Seeds, family=binomial, random.distribution="gq")
summary(mod.ml)


# Using MCMCglmm
prior <- list(R=list(V=1, nu=0.002))
mod.mcmc <- MCMCglmm(cbind(r, n-r) ~ x1*x2, family="multinomial2", data=Seeds, prior=prior)
summary(mod.mcmc$Sol)


# Using DPpackage
-----------------------------------------------
I will keep updating by adding new things (estimation procedures, predictive simulations, etc.)

1 comment:

tallertb said...

hi Shige!
I'd like to know if you've updated it. In fact,I'm working with a similar analysis and I'd like how I could contact with you. My email is:
mcasals@aspb.cat.

Thanks in advance,
Martí.

Counter