ScotCancer

The likelihood of direct CAR models and Binomial and Poisson GLM with latent CAR variables are approximated by the Monte Carlo likelihood. The Maximum Monte Carlo likelihood estimator is found either by an iterative procedure of directly maximising the Monte Carlo approximation or by a response surface design method.Reference for the method can be found in the DPhil thesis in Z. Sha (2016). For application a good reference is R.Bivand et.al (2017) <doi:10.1016/j.spasta.2017.01.002>.

Zhe Sha

mclcar

Estimating Conditional Auto-Regressive (CAR) Models using Monte
Carlo Likelihood Methods

ScotCancer function

<p>The data object for the Monte Carlo likelihood estiation and it follows the same structure as described in <code>CAR.simGLM</code>. The covX is the model matrix with intercept and the fixed effect Paff</p>

ScotCancer: Scottish lip cancer dataset from Clayton and Kaldor (1987)

Description

Arguments

Format

References