R/fit_model.R

In LudvigOlsen/cvms: Cross-Validation for Model Selection

fit_model <- function(model_specifics, train_set) {

  # Extract arguments from model_specifics
  model_formula <- model_specifics[["model_formula"]]
  model_fn <- model_specifics[["model_fn"]]
  hparams <- extract_hparams(model_specifics)

  # Some model types don't accept string formulas
  # so we convert it to an actual formula object
  if (is.character(model_formula)) {
    model_formula <- stats::as.formula(model_formula)
  }

  model_fn(
    formula = model_formula,
    train_data = train_set,
    hyperparameters = hparams
  )
}