simsList: Convert MCMC samples to a list of simulations (deprecated)

In mikemeredith/dumpJAGS: Run JAGS and Regularly Save Output to Files

Description

Takes an object of class mcmc.list and converts it to a list with one component for each parameter present. Parameters which are not scalar are stored as arrays of the appropriate dimension. Ragged arrays in the JAGS output result in all-NA columns in the output - see Examples.

We now recommend the use of mcmcOutput objects instead of simsList. This function will be removed in future versions.

Usage

simsList(mc)

Arguments

mc	an object of class mcmc.list.

Value

A list with one component for each parameter monitored. This is the same format as the sims.list component in the output of other JAGS wrappers.

Author(s)

Mike Meredith.

Examples

# Salamander occupancy again:
sal <- rep(0:4, c(21,12,1,4,1))

# JAGS code for psi(.) p(.) occupancy model but with additional parameters:
#   p appears to be a 2x2 matrix but p[1,1] is undefined;
#   this will be a ragged array in the output.
modelText <- "
model {
  for(i in 1:nSites) {
    z[i] ~ dbern(psi)
    y[i] ~ dbin(p[1,2] * z[i], n)
  }
  psi ~ dbeta(1, 1)
  p[1,2] ~ dbeta(1, 1)
  p[2,1] ~ dbeta(1, 1)
  p[2,2] <- 3
} "
writeLines(modelText, con = "JAGSmodel.txt")

JAGSdata <- list(y = sal, n = 5, nSites = length(sal))
inits <- function(chain) list(z = rep(1, 39))
wanted <- c("p", "psi")

res1 <- saveJAGS(JAGSdata, inits, wanted, "JAGSmodel.txt",
        chains=2, sample2save=1000, nSaves=4, burnin=1000, thin=1,
        fileStub="testing")
str(res1)
summary(res1)
  # p has 3 elements

# Load the results into R as an mcmc.list object
mcmc <- combineSaves(res1)
str(mcmc)
#... and convert to a sims.list:
sl <- simsList(mcmc)
str(sl)
  # sl$p is an 8000 x 2 x 2 array, with sl$p[, 1, 1] filled with NAs
apply(sl$p, 2:3, mean)

mikemeredith/dumpJAGS documentation built on March 18, 2021, 3:36 p.m.