krige (version 0.6.2)

update.krige: Update Model

Description

Update the Markov chain associated with the metropolis.krige model

Usage

# S3 method for krige
update(object, n.iter, n.burnin = 0, combine = FALSE, ...)

Arguments

object

An krige object from the metropolis.krige function.

n.iter

Number of iterations for the update run.

n.burnin

The number of burnin iterations. Defaults to 0.

combine

Logical value indicate whether to combine the update run with original output.

…

Additional arguments passed to update methods. Not supported for krige objects.

Value

An object of class krige that includes the output MCMC matrix of sampled values from the posterior distribution as well as the record of function arguments, model frame, acceptance rates, and standing parameters.

Details

Since MCMC calculations typically need to run relatively long. This function can continue the MCMC calculations by metropolis.krige(). The parameters of the original model and the estimated results from the previous run are passed through the krige object.

As geospatial estimation can be computationally taxing, the users may want to preserve more iterations of the posterior samples. The combine argument can be used to indicate whether combine the updated run with previous results. This includes both the posterior samples and acceptance rates.

Examples

# NOT RUN {
# Summarize Data
summary(ContrivedData)

# Set seed
set.seed(1241060320)

M <- 100
#M<-10000

contrived.run <- metropolis.krige(y ~ x.1 + x.2, coords = c("s.1","s.2"), 
   data = ContrivedData, n.iter = M, range.tol = 0.05)
summary(contrived.run)

# Update run
contrived.run2 <- update(contrived.run, n.iter = M, combine = TRUE)

summary(contrived.run2)
#plot(contrived.run2)
# }
# NOT RUN {
# }