Fit a spatial model with INLA

This function fits a spatial model using Gaussian random fields with INLA. The function is not intended to be used directly by the user, but documentation is included so that you can see the argument values that you might pass from fit_survey_sets(). It is called internally by fit_survey_sets().

fit_inla(dat, response = "present", n_knots = 50,
  family = "binomial", max_edge = c(20, 100), kmeans = FALSE,
  plot = FALSE, fit_model = TRUE, extend = list(n = 8, offset =
  -0.1), offset = c(5, 25), cutoff = 10, include_depth = TRUE,
  verbose = FALSE, debug = FALSE, trials = 10)

Arguments

dat	A data frame from scale_survey_predictors().
response	A character value representing the column name for the response variable.
n_knots	Number of knots in the mesh. Passed to stats::kmeans().
family	Family to pass on to INLA. Like we should be "binomial" or "gamma".
max_edge	Value to pass on to INLA::inla.mesh.2d().
kmeans	Logical for whether to use stats::kmeans() to calculate the knot locations before generating the mesh with INLA::inla.mesh.create() (if TRUE) or use INLA::inla.mesh.2d() if FALSE.
plot	Logical for whether the INLA mesh should be plotted.
fit_model	Logic for whether the model should be fitted. Useful for debugging. Will return NA in the model component if FALSE.
extend	Value to pass on to INLA::inla.mesh.create().
offset	Value to pass on to INLA::inla.mesh.2d().
cutoff	Value to pass on to INLA::inla.mesh.2d().
include_depth	Logical for whether to include depth as a predictor.
verbose	Logical for verbose output from INLA.
debug	Logical for debugging.
trials	Number of times to retry the model if it fails to fit.