ml_tune: Function to tune ML algos for multiple time series...
In vidarsumo/sumots: Helper functions for Sumo Analytics

ml_tune() will tune up to nine different ML algorithms used in forecasting

ml_tune(
  parsnip_recipe,
  modeltime_recipe,
  recipe_spec_catlight,
  vfold,
  grid_size,
  cv_repeats,
  parallel_type = "everything",
  return = c("modellist", "modeltable", "both"),
  splits,
  models = c("xgboost", "rf", "cubist", "svm_rbf", "svm_poly", "glmnet", "knn", "mars",
    "prophet_boost", "lightgbm", "catboost"),
  learn_rate = NULL,
  min_n = NULL,
  tree_depth = NULL,
  loss_reduction = NULL,
  save_modeltime_table = TRUE,
  include_simple_model_ensemble = FALSE,
  top_ensemble = NULL,
  GPU = FALSE
)

`parsnip_recipe`	Recipe for parsnip models
`modeltime_recipe`	Recipe for modeltime models
`recipe_spec_catlight`	Recipes for catboost and lightgbm
`vfold`	Number of folds used in K-fold cross validation
`grid_size`	The size of the hyperparameter grid used for tuning the parameters
`cv_repeats`	How many CV repeats to use
`return`	What do you want to return. List of workflows, modeltime table or both?
`splits`	The split object to use
`models`	Choose which models to use. Choose any combination of xgboost, rf, cubist, svm_rbf, svm_poly, glmnet, knn, mars or prophet_boost
`learn_rate`	Upper and lower bound of learning rate to try out during tuning. NULL equals default values from the dials package
`min_n`	Upper and lower bound of min_n to try out during tuning. NULL equals default values from the dials package
`tree_depth`	Upper and lower bound of tree_depth to try out during tuning. NULL equals default values from the dials package
`loss_reduction`	Upper and lower bound of loss_reduction to try out during tuning. NULL equals default values from the dials package
`save_modeltime_table`	Should modeltime table be saved after tuning each algorithm. Defaults to TRUE
`include_simple_model_ensemble`	Should simple average of all the models be included? Defaults to TRUE
`top_ensemble`	Creates an ensemble only from the top models. Defaults to 3. Set = NULL if you don't want to use.
`GPU`	Should GPU be used where applicable? Defaults to FALSE
`parallel_over`	A single string containing either "resamples" or "everything" describing how to use parallel processing. See ?control_grid