SUMMER (version 0.2.3)

fitINLA: Fit INLA models to direct estimators with a yearly model.

Description

Fit INLA models to direct estimators with a yearly model.

Usage

fitINLA(data, Amat, geo, formula = NULL, rw = 2, is.yearly = TRUE,
  year_names, year_range = c(1980, 2014), m = 5, na.rm = TRUE,
  redo.prior = FALSE, priors = NULL, type.st = 1, useHyper = FALSE,
  a.iid = NULL, b.iid = NULL, a.rw1 = NULL, b.rw1 = NULL,
  a.rw2 = NULL, b.rw2 = NULL, a.icar = NULL, b.icar = NULL,
  options = list(dic = T, mlik = T, cpo = T, openmp.strategy =
  "default"), verbose = FALSE)

Arguments

data

Combined dataset

Amat

Adjacency matrix for the regions

geo

Geo file

formula

INLA formula. Defaults to RW2, ICAR, IID time, IID, region, IID survey effect, IID time-region interaction, IID survey-region interaction, and IID survey-time-region interaction.

rw

Take values 1 or 2, indicating the order of random walk.

is.yearly

Logical indicator for fitting yearly or period model.

year_names

string vector of year names

year_range

Entire range of the years (inclusive) defined in year_names.

m

Number of years in each period.

na.rm

Logical indicator of whether to remove rows with NA values in the data. Default set to TRUE.

redo.prior

Logical indicator of whether to re-estimate hyperparameters

priors

priors from simhyper

type.st

type for space-time interaction

useHyper

option to manually set all hyperpriors

a.iid

hyperparameter for i.i.d random effects, only need if useHyper = TRUE

b.iid

hyperparameter for i.i.d random effects, only need if useHyper = TRUE

a.rw1

hyperparameter for RW1 random effects, only need if useHyper = TRUE

b.rw1

hyperparameter for RW1 random effects, only need if useHyper = TRUE

a.rw2

hyperparameter for RW2 random effects, only need if useHyper = TRUE

b.rw2

hyperparameter for RW2 random effects, only need if useHyper = TRUE

a.icar

hyperparameter for ICAR random effects, only need if useHyper = TRUE

b.icar

hyperparameter for ICAR random effects, only need if useHyper = TRUE

options

list of options to be passed to control.compute() in the inla() function.

verbose

logical indicator to print out detailed inla() intermediate steps.

Value

INLA model fit using the provided formula, country summary data, and geographic data

See Also

countrySummary

Examples

# NOT RUN {
data(DemoData)
data(DemoMap)
years <- levels(DemoData[[1]]$time)

# obtain direct estimates
data <- countrySummary_mult(births = DemoData, 
years = years, 
regionVar = "region", timeVar = "time", 
clusterVar = "~clustid+id", 
ageVar = "age", weightsVar = "weights", 
geo.recode = NULL)

# obtain maps
geo <- DemoMap$geo
mat <- DemoMap$Amat

# Simulate hyperpriors
priors <- simhyper(R = 2, nsamp = 1e+05, nsamp.check = 5000, Amat = mat, only.iid = TRUE)

# combine data from multiple surveys
data <- aggregateSurvey(data)

# Model fitting with INLA
years.all <- c(years, "15-19")
fit <- fitINLA(data = data, geo = geo, Amat = mat, 
year_names = years.all, year_range = c(1985, 2019),
 priors = priors, rw = 2,
 is.yearly=TRUE, m = 5, type.st = 4)
# Projection
out <- projINLA(fit, Amat = mat, is.yearly = TRUE)
plot(out, is.yearly=TRUE, is.subnational=TRUE) + ggplot2::ggtitle("Subnational yearly model")

# }
# NOT RUN {
# }