GeoAdjust (version 2.0.1)

estimateModel: Estimates model parameters

Description

Estimates model parameters

Usage

estimateModel(
  data = NULL,
  options = NULL,
  priors = NULL,
  n.sims = NULL,
  log_tau = NULL,
  log_kappa = NULL,
  USpatial = 1,
  alphaSpatial = 0.05,
  UNugget = 1,
  alphaNug = 0.05
)

Value

Model estimation results of class GAmodel. The output consists of four elements: A data frame containing the estimated model parameters and the corresponding 95

from autodifferentiation of TMB, A matrix containing the sampled coefficient effect sizes and the random effect coefficients, A character string indicating the likelihood type in the model.

Arguments

data

A data input list that is created by prepareInput() function.

options

A list containing two components, namely, random and covariates. The function accounts for jittering both in the spatial random effect and in covariates (if there are any) by default. However, the jittering adjustment can be turned off in either the random effect or in covariates, or both, by setting the corresponding component of the list to zero.

priors

A list of two components. Beta is a vector of two elements and passes the parameters of the Gaussian prior that will be assigned to the covariates (including the intercept). The first element of it is the mean and the second one is the standard deviation of Gaussian prior. Range is a value representing the median range in kilometers, which will be used for constructing the PC (Penalized-complexity) priors.

n.sims

number of samples to be drawn for each model parameter

log_tau

SPDE parameter related to the spatial precision

log_kappa

SPDE parameter related to the range and spatial precision

USpatial

The threshold that is crossed by the the variance prior.

alphaSpatial

The probability of crossing the threshold for the variance prior.

UNugget

The threshold that is crossed by the prior on the nugget standard deviation. It will only be used when the likelihood is Gaussian.

alphaNug

The probability of crossing the threshold for the prior on the nugget standard deviation. It will only be used when the likelihood is Gaussian.

Examples

# \donttest{
path1 <- system.file("extdata", "exampleInputData.rda", package = "GeoAdjust")
path2 <- system.file("extdata", "exampleMesh.rda", package = "GeoAdjust")
load(path1)
load(path2)
results <- estimateModel(data = exampleInputData, priors = list(beta = c(0,1),
range = 114), USpatial = 1, alphaSpatial = 0.05, UNugget = 1, alphaNug = 0.05, n.sims = 1000)
# }