Auto Multiple-Seasonal ARIMA Forecast Learner

Automatic order selection for Multiple-Seasonal ARIMA. Picks orders minimising the chosen information criterion. Calls smooth::auto.msarima() from package smooth.

Dictionary

This mlr3::Learner can be instantiated via the dictionary mlr3::mlr_learners or with the associated sugar function mlr3::lrn():

mlr_learners$get("fcst.auto_msarima")
lrn("fcst.auto_msarima")

Meta Information

• Task type: “fcst”

• Predict Types: “response”

• Feature Types: “logical”, “integer”, “numeric”, “character”, “factor”, “ordered”, “POSIXct”, “Date”

• Required Packages: mlr3, mlr3forecast, smooth

Parameters

Id	Type	Default	Levels
orders	untyped	list(ar = c(3, 3), i = c(2, 1), ma = c(3, 3))
lags	untyped	-
initial	character	backcasting	backcasting, optimal, two-stage, complete
ic	character	AICc	AICc, AIC, BIC, BICc
loss	character	likelihood	likelihood, MSE, MAE, HAM, MSEh, TMSE, GTMSE, MSCE, GPL
holdout	logical	FALSE	TRUE, FALSE
bounds	character	usual	usual, admissible, none
silent	logical	TRUE	TRUE, FALSE
regressors	character	use	use, select, adapt

References

Svetunkov I (2023). “Smooth forecasting with the smooth package in R.” 2301.01790, https://arxiv.org/abs/2301.01790.

Svetunkov, Ivan (2023). Forecasting and Analytics with the Augmented Dynamic Adaptive Model (ADAM), 1st edition. Chapman and Hall/CRC. doi:10.1201/9781003452652 . https://openforecast.org/adam/.

See also

• Chapter in the mlr3book: https://mlr3book.mlr-org.com/chapters/chapter2/data_and_basic_modeling.html#sec-learners

• Package mlr3learners for a solid collection of essential learners.

• Package mlr3extralearners for more learners.

• Dictionary of Learners: mlr3::mlr_learners

• as.data.table(mlr_learners) for a table of available Learners in the running session (depending on the loaded packages).

• mlr3pipelines to combine learners with pre- and postprocessing steps.

• Package mlr3viz for some generic visualizations.

• Extension packages for additional task types:

  • mlr3proba for probabilistic supervised regression and survival analysis.

  • mlr3cluster for unsupervised clustering.

• mlr3tuning for tuning of hyperparameters, mlr3tuningspaces for established default tuning spaces.

Other Learner: LearnerFcst, mlr_learners_fcst.adam, mlr_learners_fcst.arfima, mlr_learners_fcst.arima, mlr_learners_fcst.auto_adam, mlr_learners_fcst.auto_arima, mlr_learners_fcst.auto_ces, mlr_learners_fcst.auto_gum, mlr_learners_fcst.bagged, mlr_learners_fcst.bats, mlr_learners_fcst.ces, mlr_learners_fcst.croston, mlr_learners_fcst.ets, mlr_learners_fcst.gum, mlr_learners_fcst.holt_winters, mlr_learners_fcst.mean, mlr_learners_fcst.msarima, mlr_learners_fcst.nnetar, mlr_learners_fcst.prophet, mlr_learners_fcst.random_walk, mlr_learners_fcst.sma, mlr_learners_fcst.spline, mlr_learners_fcst.stlm, mlr_learners_fcst.struct_ts, mlr_learners_fcst.tbats, mlr_learners_fcst.theta, mlr_learners_fcst.tscount, mlr_learners_fcst.tslm

Super classes

mlr3::Learner -> mlr3::LearnerRegr -> LearnerFcst -> LearnerFcstAutoMsarima

Methods

Public methods

• LearnerFcstAutoMsarima$new()

• LearnerFcstAutoMsarima$clone()

Inherited methods

• mlr3::Learner$base_learner()
• mlr3::Learner$configure()
• mlr3::Learner$encapsulate()
• mlr3::Learner$format()
• mlr3::Learner$help()
• mlr3::Learner$predict()
• mlr3::Learner$predict_newdata()
• mlr3::Learner$print()
• mlr3::Learner$reset()
• mlr3::Learner$selected_features()
• mlr3::Learner$train()
• mlr3::LearnerRegr$predict_newdata_fast()

---

LearnerFcstAutoMsarima$new()

Creates a new instance of this R6 class.

Usage

LearnerFcstAutoMsarima$new()

---

LearnerFcstAutoMsarima$clone()

The objects of this class are cloneable with this method.

Usage

LearnerFcstAutoMsarima$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

# Define the Learner and set parameter values
learner = lrn("fcst.auto_msarima")
print(learner)
#> 
#> ── <LearnerFcstAutoMsarima> (fcst.auto_msarima): Auto Multiple-Seasonal ARIMA ──
#> • Model: -
#> • Parameters: list()
#> • Packages: mlr3, mlr3forecast, and smooth
#> • Predict Types: [response]
#> • Feature Types: logical, integer, numeric, character, factor, ordered,
#> POSIXct, and Date
#> • Encapsulation: none (fallback: -)
#> • Properties: featureless and missings
#> • Other settings: use_weights = 'error', predict_raw = 'FALSE'

# Define a Task
task = tsk("airpassengers")

# Create train and test set
ids = partition(task)

# Train the learner on the training ids
learner$train(task, row_ids = ids$train)

# Print the model
print(learner$model)
#> Time elapsed: 0.42 seconds
#> Model estimated using auto.adam() function: SARIMA(0,1,1)[1](0,1,1)[12]
#> With backcasting initialisation
#> Distribution assumed in the model: Normal
#> Loss function type: likelihood; Loss function value: 346.9397
#> ARMA parameters of the model:
#>         Lag 1  Lag 12
#> MA(1) -0.2176 -0.1701
#> 
#> Sample size: 96
#> Number of estimated parameters: 3
#> Number of degrees of freedom: 93
#> Information criteria:
#>      AIC     AICc      BIC     BICc 
#> 699.8793 700.1402 707.5724 708.1677 

# Importance method
if ("importance" %in% learner$properties) print(learner$importance())

# Make predictions for the test rows
predictions = learner$predict(task, row_ids = ids$test)

# Score the predictions
predictions$score()
#> regr.mse 
#> 740.7169