knmodels.sample: Spacetime interaction models sampler function

In inbo/INLA: Full Bayesian Analysis of Latent Gaussian Models using Integrated Nested Laplace Approximations

Description

It implements the sampling method for the models in Knorr-Held, L. (2000) considering the algorithm 3.1 in Rue & Held (2005) book.

Usage

inla.knmodels.sample(
  graph,
  m,
  type=4,
  intercept=0, 
  tau.t=1,
  phi.t=0.7,
  tau.s=1,
  phi.s=0.7,
  tau.st=1,
  ev.t=NULL,
  ev.s=NULL)

Arguments

graph
m	Time dimention.
type	Integer from 1 to 4 to identify one of the four interaction type.
intercept	A constant to be added to the linear predictor
tau.t	Precision parameter for the main temporal effect.
phi.t	Mixing parameter in the bym2 model assumed for the main temporal effect.
tau.s	Precision parameter for the main spatial effect.
phi.s	Mixing parameter in the bym2 model assumed for the main spatial effect.
tau.st	Precision parameter for the spacetime effect.
ev.t	Eigenvalues and eigenvectors of the temporal precision matrix structure.
ev.s	Eigenvalues and eigenvectors of the spatial precision matrix structure.

Value

A list with the following elements

time	The time index for each obervation, with length equals m*n.
space	The spatial index for each obervation, with length equals m*n.
spacetime	The spacetime index for each obervation, with length equals m*n.
x	A list with the following elements
t.iid	The unstructured main temporal effect part.
t.str	The structured main temporal effect part.
t	The main temporal effect with length equals 2m.
s.iid	The unstructured main spatial effect part.
s.str	The structured main spatial effect part.
s	The main spatial effect with length equals 2n.
st	The spacetime interaction effect with length equals m*n.
eta	The linear predictor with length equals n*m.

Author(s)

Elias T. Krainski

See Also

inla.knmodels for model fitting

inbo/INLA documentation built on Dec. 6, 2019, 9:51 a.m.