Retrieve predictions for stdglm objects

Extracts out-of-sample predictions from an stdglm object.

Usage

# S3 method for class 'stdglm'
predict(object, type = "response_mat", Coo_sf_pred = NULL, ...)

Arguments

object

An stdglm object.

type

A character string indicating the type of coefficients. Options are:

• response_mat: Posterior mean of the predictive distribution for the response variable. Returns a p_new-by-t_new matrix (default).

• response_df: Mean, standard deviation and 95 % credible interval for the response variable (default). Returns a dataframe.

• tvc: Mean, standard deviation and 95 % credible interval for the temporal effects of varying coefficients.

• svc: Mean, standard deviation and 95 % credible interval for the spatial effects of varying coefficients.

• stvc: Mean, standard deviation and 95 % credible interval for the spatio-temporal effects of varying coefficients.

Coo_sf_pred

A simple feature object from package sf with the prediction points, whose geometry is used for spatial effects (optional).

...

Additional arguments (currently ignored).

Value

Either a matrix or a dataframe or an sf object with posterior mean and 95% credible interval bounds. The function returns NULL if predictions are not available in the stdglm object.

Details

Returns the posterior mean of the predictive distribution and the associated 95% credible intervals for the out-of-sample data points, and for the random variable specified in the input type.

See also

stdglm, sf

Examples

if (FALSE) { # \dontrun{
# Assuming `mod` is a fitted stdglm object
predictions <- predict(mod)
colMeans(predictions) # Get the average predictions across all locations
pred_df <- predict(mod, type = 'response_df')
head(pred_df)
} # }