trafoensemble {deeptrafo}R Documentation

Transformation ensembles

Description

Transformation ensembles

Usage

trafoensemble(
  formula,
  data,
  n_ensemble = 5,
  verbose = FALSE,
  print_members = TRUE,
  stop_if_nan = TRUE,
  save_weights = TRUE,
  callbacks = list(),
  save_fun = NULL,
  seed = seq_len(n_ensemble),
  tf_seeds = seq_len(n_ensemble),
  ...
)

Arguments

formula

Formula specifying the response, interaction, shift terms as response | interacting ~ shifting. auto-regressive transformation models (ATMs).

data

Named list or data.frame which may contain both structured and unstructured data.

n_ensemble

Numeric; number of ensemble members to fit.

verbose

Logical; whether to print training in each fold.

print_members

Logical; print results for each member.

stop_if_nan

Logical; whether to stop ensembling if NaN values occur

save_weights

Logical; whether to save the ensemble weights.

callbacks

List; callbacks used for fitting.

save_fun

Function; function to be applied to each member to be stored in the final result.

seed

Numeric vector of length n_ensemble; seeds for model re-initialization. Changing these seeds does not change the parameters of the interacting predictor coef(obj, which_param = "interacting"), change tf_seeds to adapt those coefficients.

tf_seeds

Numeric vector of length n_ensemble; explicit seed for changing the parameters of the interacting predictor. Distinct from seed which is used for weight re-initialization of the rest of the model (i.e., the shifting predictor and potential neural network components in the interacting component).

...

Further arguments passed to deeptrafo and fit.

Value

Ensemble of "deeptrafo" models with list of training histories and fitted weights included in ensemble_results. For details see the return statment in ensemble.


[Package deeptrafo version 1.0-0 Index]