Fit multiple workflows for model stacking — h3sdm_fit_models

Fits a set of tidymodels workflows using resampling with fit_resamples() and control_stack_resamples() to generate objects suitable for stacking. For each workflow, both the resampled model (cv_model) and the original workflow are returned. This is intended as a helper function for creating ensembles of multiple models in the h3sdm workflow.

Usage

h3sdm_fit_models_stack(workflows, data_split, presence_data = NULL)

Arguments

workflows: A named list of tidymodels workflows. Each element should be a workflow object.
data_split: A resampling object, e.g., generated by rsample::vfold_cv().
presence_data: (Optional) A data frame containing the data to fit the workflows on.

Value

A list with one element models. Each element of models is itself a list with:

cv_model: the resampled model (tune_results object)
workflow: the original workflow object

Details

Fit multiple workflows with resampling for stacking

Examples

if (FALSE) { # \dontrun{
# Example requires prepared workflows and resampling object:
# wfs <- list(random_forest = my_workflow_rf, glmnet = my_workflow_glm)
# scv <- rsample::vfold_cv(my_data, v = 5)
results <- h3sdm_fit_models_stack(workflows = wfs, data_split = scv)
results$models$random_forest$cv_model
results$models$random_forest$workflow
} # }