My goal is to develop a stochastic search variable selection strategy (SSVS) coupled to a Besag-York-Molié (BYM) spatial model. In particular, adapt to NIMBLE an efficient marginalized implementation in the Stan language. The project will analyze a rich set of epidemiological data from the Ebola epidemic in West Africa between 2013 and 2016.
The models are being further developed and wrapped in a R package. Better mixing conditions for the MCMC has already been achieved and fake data generation has been standardized. All the work will be properly linked here soon.