ComputePostPredDist: Computing the posterior predictive distribution of total area...
In PointProcess/SealPupProduction-JRSSC-code:

Computes the posterior predictive distribution of total area counts based on a set of samples from the posterior field. The function first splits the samples into several sublists and saves them to disk. Then computes the eta for each point in each grid along with the sd and so on. Finally computes the dist of the mean of the Poisson counts corresponding to the posterior predictive dist of total area counts.

ComputePostPredDist(
  samp,
  spatial,
  parallelize.noSplits = parallelize.numCores,
  parallelize.numCores,
  tempFolder,
  use.covariates,
  additional.iid.term,
  covariate.fitting,
  gridList,
  covGridList,
  nxy,
  poisson.maxEvals,
  noMeshPoints,
  extraNonlinear.covGridList,
  delete.samp = T
)

`samp`	List with all posterior samples, one sample in each sublist (as obtained from inla.posterior.sample)
`spatial`	Logical, indicating whether the model fitted includes a spatial spde term
`parallelize.noSplits`	Numeric, deciding how many sublists samp should be splitted into. Should be a multiple of parallelize.numCores for the highest efficiency. The larger number the less memory is used (and longer time)
`parallelize.numCores`	Numeric, corresponding to the number of cores any parallelization should be run at
`tempFolder`	Path to the folder where temporary files (samp subfiles and gridded eta subfiles)
`use.covariates`	Logical, indicating whether covariates are used or not (see description!)
`additional.iid.term`	Logical, indicating whether to include an additional iid (Gaussian) term in the latent field specification. FALSE is default
`covariate.fitting`	String, indicating how to model covariates. "linear", quadratic (default) or "linAndLog", or FALSE for no covariates
`gridList`	List containing information about the grid, being the output of the function GridCreation
`covGridList`	List containing information about the covariates at the grid (also when use.covariates = FALSE), being the output of the function covAtNewGrid
`nxy`	Numberic vector of size 2, giving the dimension in x- and y-direction for the grid
`poisson.maxEvals`	Numeric, corresponding to maximum number of points the Poisson distribution should be evaluated at (a much smaller number is typically used)
`noMeshPoints`	Numeric, corresponding to the number of points in the mesh being used
`extraNonlinear.covGridList`	List of additional projection objects related to the nonlinear covariate effect when applicable

List containing the resulting posterior predicitve distribution for all evaluation points (also returned) in addition to the mean field and sd field computed from the samples)

PointProcess/SealPupProduction-JRSSC-code documentation built on Jan. 27, 2020, 10:06 p.m.