augment_inla: Run 'inla' with new data and give list of new predictors'...
In FrankLef/eflINLA: Utilities used with INLA by FL

Description Usage Arguments Details Value Examples

Run inla with new data and give list of new predictors' positions.

1	augment_inla(.result, newdata)

`.result`	`inla` object.
`newdata`	Dataframe of new data. The column with the response (outcome, dependent variable) must be filled with `NA`s.

Run inla with new data to allow the use of summary.linear.predictor, summary.fitted.values, marginals.linear.predictor, etc. Then put it in a list with the positions of the newdata described just below. This function is necessary because INLA always keep the original data, as a result we constantly use the row position of the new data when extracting linear predictors, expected predictors (fitted values), etc. More specifically, this function will

merge data with newdata
create a vector with the newdata's position in the inla data
set control.predictor = list(compute = TRUE)
set offset <- NULL to avoid error message
set control.compute = list(config = TRUE), otherwise error from INLA::inla.posterior.sample.

List with the following items

inla: updated inla object.
new_pos: integer() of newdata positions.

the_data <- data.frame(
 x1 = rnorm(20, mean = 1, sd = 0.5),
 x2 = rnorm(20, mean = 2, sd = 1),
 sigma = rexp(20))
the_data$mu <- the_data$x1 + the_data$x2
the_data$y <- rnorm(nrow(the_data), mean = the_data$mu, sd = the_data$sigma)
the_inla <- INLA::inla(formula = y ~ x1 + x2, data = the_data)
# the new data used for fitted/predicted values
the_newdata <- data.frame(
 x1 = modelr::seq_range(the_data$x1, n = 5),
 x2 = modelr::seq_range(the_data$x2, n = 5))
augment_inla(the_inla, newdata = the_newdata)