Fit an INLA model to one cv_set and get prediction intervals

Usage

fit_inla(
  cv_set,
  y_var,
  pred_vars,
  id_vars,
  reff_var = NULL,
  hyper_priors = list(prec.unstruct = c(1, 5e-04), prec.spatial = c(1, 5e-04),
    prec.timerw1 = c(1, 0.01)),
  return_model = FALSE,
  quantile_levels = c(0.01, 0.025, seq(0.05, 0.95, by = 0.05), 0.975, 0.99),
  sample_pi = FALSE,
  W_orgUnit = NULL,
  verbose = FALSE
)

Arguments

cv_set: list object containing analysis and assessment data.frames
y_var: string name of column of observed values
pred_vars: vector string of predictor variables to use in model
id_vars: names of columns used as keys/ids in data.frames
reff_var: string or formula for random effects
hyper_priors: priors for spatial and temporal random effects
return_model: whether or not to return the INLA model in addition to predictions
quantile_levels: quantile levels to use when predicted via quantile regression. Default = c(0.01,0.025, seq(0.05,0.95, by = 0.05), 0.975, 0.99)
sample_pi: whether to estimate the prediction intervals by sampling the posterior. Default (FALSE) uses a trick to put all error into an iid random effect.
W_orgUnit: graph of orgUnit to use in spatial structure. If NULL (default), will not fit a spatial structure.
verbose: whether to run the INLA model verbosely. Default = FALSE

Value

dataframe of predictions intervals