Skip to contents

Create meta_learner predictions from the list of BART fit objects and predictions of base learners

Source: R/meta_learner.R
meta_learner_predict.Rd

The meta learner used in this package, Bayesian Additive Regression Tree (BART), is not explicitly a spatiotemporal model, but the input covariates (outputs of each base learner) are S-T based.

Usage

meta_learner_predict(meta_fit, base_outputs_stdt, nthreads = 2)

Arguments

meta_fit

list of BART objects from meta_learner_fit

base_outputs_stdt

stdt object. list with datatable containing lat, lon, time and the covariates (outputs of each base learner) at prediction locations and crs.

nthreads

integer(1). Number of threads used in BART::predict.wbart

Value

meta_pred: the final meta learner predictions

Note

The predictions can be a rast or sf, which depends on the same respective format of the covariance matrix input - cov_pred

References

https://rspatial.github.io/terra/reference/predict.html

Examples

NULL
#> NULL