tune_deepar: Function to tune DeepAR
In vidarsumo/sumots: Helper functions for Sumo Analytics

Function to tune DeepAR

tune_deepar(
  id,
  freq,
  recipe,
  horizon,
  splits,
  length,
  cv_slice_limit,
  most_important = TRUE,
  assess = "12 weeks",
  skip = "4 weeks",
  initial = "12 months",
  multiple_gpu = FALSE,
  no_gpu,
  min_obs_cv_train = 1,
  clip_gradient = 10,
  epochs = NULL,
  lookback = NULL,
  batch_size = NULL,
  learn_rate = NULL,
  id_use,
  num_cells = NULL,
  num_layers = NULL,
  scale = NULL,
  dropout = NULL
)

`id`	A quoted column name that tracks the GluonTS FieldName "item_id"
`freq`	A pandas timeseries frequency such as "5min" for 5-minutes or "D" for daily.
`recipe`	A gluonts recipe
`horizon`	The forecast horizon
`length`	The number of distinct hyperparameter for each tunable parameter
`cv_slice_limit`	How many slice/folds in the tsCV
`assess`	The number of samples used for each assessment resample
`skip`	A integer indicating how many (if any) additional resamples to skip to thin the total amount of data points in the analysis resample.
`initial`	The number of samples used for analysis/modeling in the initial resample.
`multiple_gpu`	Should more than one GPU be used
`no_gpu`	How many, if more than one, should be used
`min_obs_cv_train`	Minimum observation in the training set during cross validation
`clip_gradient`	Maximum value of gradient. The gradient is clipped if it is too large (default: 10)
`epochs`	Number of epochs. Importance 1 of 7
`lookback`	Lookback length. If NULL, will be randomly chosen. Importance 2 of 7
`batch_size`	batch_size Number of examples in each batch. Importance 3 of 7
`learn_rate`	Learning rate. Importance 4 of 7
`id_use`	ID used during training
`num_cells`	Number of RNN cells for each layer. Importance 5 of 7
`num_layers`	Number of RNN layers. No info on importance
`scale`	Scales numeric data by id group using mean = 0, standard deviation = 1 transformation. No info on importance