trainer_params: General Tuning Parameters for GluonTS Trainers (All Models)

In business-science/modeltime.gluonts: 'GluonTS' Deep Learning

General Tuning Parameters for GluonTS Trainers (All Models)

Description

These parameters are part of the gluonts.mx.trainer.Trainer() function.

Usage

lookback_length(range = c(dials::unknown(), dials::unknown()), trans = NULL)

num_batches_per_epoch(range = c(2, 100), trans = NULL)

learn_rate_decay_factor(range = c(0, 1), trans = NULL)

scale_values(values = c(TRUE, FALSE))

Arguments

range	A two-element vector holding the defaults for the smallest and largest possible values, respectively. If a transformation is specified, these values should be in the transformed units.
trans	A trans object from the scales package, such as scales::log10_trans() or scales::reciprocal_trans(). If not provided, the default is used which matches the units used in range. If no transformation, NULL.
values	A character string of possible values.

Details

The main parameters for GluonTS Model Training are:

• epochs: The number of iterations of training. See dials::epochs().

• num_batches_per_epoch: Number of batches at each epoch.

• learn_rate: The rate at which learning is changed. See dials::learn_rate().

• learn_rate_decay_factor: Factor by which to decrease the learning rate.

• penalty: The weight decay (or L2 regularization) coefficient. See dials::penalty().

Additional useful parameters that have been added to Modeltime are:

• lookback_length: Number of time units that condition the predictions.

• scale_values: Scales numeric data by ‘id’ group using mean = 0, standard deviation = 1 transformation.

business-science/modeltime.gluonts documentation built on Jan. 20, 2024, 3:59 a.m.