SpatialEpi: MCMC_simulation – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

MCMC_simulation

MCMC simulation to sample configurations

MCMC simulation to estimate prior and posterior quantities by sampling configurations.

MCMC_simulation(n_sim, pattern, theta_init, overlap, cluster_coords, 
p_moves_orig, J, lkhd_z, lambda)

`n_sim`	number of MCMC iterations
`pattern`	alternating pattern between unif and prop prior on single zones
`theta_init`	initial configuration
`overlap`	output of `create_geo_objects`: a list with two elements: `1. presence` which lists for each areas all the single zones it is present in and `2. cluster_list` for each single zone its component areas
`cluster_coords`	output of `create_geo_objects`: `n.zones x 2` matrix of the centering and radial area of each single zone
`p_moves_orig`	probability of sampling each of the 5 possible moves to explore sample space
`J`	maximum number of clusters/anti-clusters to consider
`lkhd_z`	values associated with each single zone to use in Metropolis-Hastings ratio
`lambda`	lambda from definition of prior on single zones

`sample`	sampled configurations
`move_trace`	trace of moves (1 = growth, 2 = trim, 3 = recenter, 4 = death, 5 = birth)
`accpt_trace`	trace of acceptance (0 = not accepted)
`ratio_trace`	trace of Metropolis-Hastings ratio

Albert Y. Kim

Wakefield J. and Kim A.Y. (2013) A Bayesian model for cluster detection. Biostatistics, 14, 752–765.