disaggregation (version 0.1.4)

predict.disag_model: Predict mean and uncertainty from the disaggregation model result

Description

predict.disag_model function takes a disag_model object created by disaggregation::disag_model and predicts mean and uncertainty maps.

Usage

# S3 method for disag_model
predict(object, newdata = NULL, predict_iid = FALSE, N = 100, CI = 0.95, ...)

Value

An object of class disag_prediction which consists of a list of two objects:

mean_prediction

List of:

• prediction Raster of mean predictions based.

• field Raster of the field component of the linear predictor.

• iid Raster of the iid component of the linear predictor.

• covariates Raster of the covariate component of the linear predictor.

uncertainty_prediction:

List of:

• realisations RasterStack of realisations of predictions. Number of realisations defined by argument N.

• predictions_ci RasterStack of the upper and lower credible intervals. Defined by argument CI.

Arguments

object

disag_model object returned by disag_model function.

newdata

If NULL, predictions are made using the data in model_output. If this is a raster stack or brick, predictions will be made over this data.

predict_iid

logical. If TRUE, any polygon iid effect from the model will be used in the prediction. Default FALSE.

N

Number of realisations. Default: 100.

CI

Confidence interval to be calculated from the realisations. Default: 0.95.

...

Further arguments passed to or from other methods.

Details

To predict over a different spatial extent to that used in the model, a RasterStack covering the region to make predictions over is passed to the argument newdata. If this is not given predictions are made over the data used in the fit.

The predict_iid logical flag should be set to TRUE if the results of the iid effect from the model are to be used in the prediction.

For the uncertainty calculations, the number of the realisations and the size of the confidence interval to be calculated are given by the arguments N and CI respectively.

Examples

if (FALSE) {
predict(fit_result)
}