Regression Bayesian Treed Linear Model Learner

Bayesian treed linear model regression. Calls tgp::btlm() from tgp.

Factor features are one-hot encoded with reference encoding before fitting. If factors are present, basemax is set to the number of non-factor features so that tree proposals account for the numeric part of the design.

Initial parameter values

• verb is initialized to 0 to silence printing.

• pred.n is initialized to FALSE to skip prediction during training.

Dictionary

This Learner can be instantiated via lrn():

lrn("regr.btlm")

Meta Information

• Task type: “regr”

• Predict Types: “response”, “se”

• Feature Types: “integer”, “numeric”, “factor”

• Required Packages: mlr3, mlr3extralearners, tgp

Parameters

Id	Type	Default	Levels	Range
bprior	character	bflat	b0, b0not, bflat, bmle, bmznot, bmzt	-
BTE	untyped	c(2000L, 7000L, 2L)		-
Ds2x	logical	FALSE	TRUE, FALSE	-
improv	logical	FALSE	TRUE, FALSE	-
itemps	untyped	NULL		-
krige	logical	TRUE	TRUE, FALSE	-
m0r1	logical	TRUE	TRUE, FALSE	-
meanfn	character	linear	constant, linear	-
pred.n	logical	-	TRUE, FALSE	-
R	integer	1		\([1, \infty)\)
trace	logical	FALSE	TRUE, FALSE	-
tree	untyped	c(0.5, 2)		-
verb	integer	-		\([0, 4]\)
zcov	logical	FALSE	TRUE, FALSE	-

References

Gramacy RB (2007). “tgp: An R Package for Bayesian Nonstationary, Semiparametric Nonlinear Regression and Design by Treed Gaussian Process Models.” Journal of Statistical Software, 19(9), 1–46. doi:10.18637/jss.v019.i09 .

Gramacy RB, Taddy M (2010). “Categorical Inputs, Sensitivity Analysis, Optimization and Importance Tempering with tgp Version 2, an R Package for Treed Gaussian Process Models.” Journal of Statistical Software, 33(6), 1–48. doi:10.18637/jss.v033.i06 .

See also

• 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).

• Chapter in the mlr3book: https://mlr3book.mlr-org.com/basics.html#learners

• mlr3learners for a selection of recommended learners.

• mlr3cluster for unsupervised clustering learners.

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

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

Author

awinterstetter

Super classes

mlr3::Learner -> mlr3::LearnerRegr -> LearnerRegrBtlm

Methods

Public methods

• LearnerRegrBtlm$new()

• LearnerRegrBtlm$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()

---

Method new()

Creates a new instance of this R6 class.

Usage

LearnerRegrBtlm$new()

---

Method clone()

The objects of this class are cloneable with this method.

Usage

LearnerRegrBtlm$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

# Define the Learner
learner = lrn("regr.btlm")
print(learner)
#> 
#> ── <LearnerRegrBtlm> (regr.btlm): Bayesian Treed Linear Model ──────────────────
#> • Model: -
#> • Parameters: pred.n=FALSE, verb=0
#> • Packages: mlr3, mlr3extralearners, and tgp
#> • Predict Types: [response] and se
#> • Feature Types: integer, numeric, and factor
#> • Encapsulation: none (fallback: -)
#> • Properties:
#> • Other settings: use_weights = 'error'

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

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

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

print(learner$model)
#> $model
#> 
#> This is a 'tgp' class object.  
#> It is basically a list with the following entries:
#> 
#>  [1] X        n        d        Z        nn       Xsplit   BTE      R       
#>  [9] linburn  g        dparams  itemps   bimprov  ess      gpcs     response
#> [17] improv   parts    trees    posts    params   m0r1    
#> 
#> See ?btgp for an explanation of the individual entries.  
#> See plot.tgp and tgp.trees for help with visualization.
#> 
#> The $trace field, if it exists, is of class 'tgptraces' 
#> and has its own print statement
#> 
#> 
#> $factor_levels
#> list()
#> 
#> $column_names
#>  [1] "am"   "carb" "cyl"  "disp" "drat" "gear" "hp"   "qsec" "vs"   "wt"  
#> 


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

# Score the predictions
predictions$score()
#> regr.mse 
#> 31.32846