dot-process_models_or_automl function

Do basic validation and transform object to a "standardized" list containing models, and their properties such as x, y, whether it is a (multinomial) clasification or not etc.

Do basic validation and transform object to a "standardized" list containing models, and their properties such as x, y, whether it is a (multinomial) clasification or not etc.

.process_models_or_automl( object, newdata, require_single_model = FALSE, require_multiple_models = FALSE, top_n_from_AutoML = NA, only_with_varimp = FALSE, best_of_family = FALSE, require_newdata = TRUE, check_x_y_consistency = TRUE )

Arguments

  • object: Can be a single model/model_id, vector of model_id, list of models, H2OAutoML object
  • newdata: An H2OFrame with the same format as training frame
  • require_single_model: If true, make sure we were provided only one model
  • require_multiple_models: If true, make sure we were provided at least two models
  • top_n_from_AutoML: If set, don't return more than top_n models (applies only for AutoML object)
  • only_with_varimp: If TRUE, return only models that have variable importance
  • best_of_family: If TRUE, return only the best of family models; if FALSE return all models in object
  • require_newdata: If TRUE, require newdata to be specified; otherwise allow NULL instead, this can be used when there is no need to know if the problem is (multinomial) classification.
  • check_x_y_consistency: If TRUE, make sure that when given a list of models all models have the same X and y. Defaults to TRUE.

Returns

a list with the following names leader, is_automl, models, is_classification, is_multinomial_classification, x, y, model

  • Maintainer: Tomas Fryda
  • License: Apache License (== 2.0)
  • Last published: 2024-01-11