GLM with Elastic Net Regularization Regression Learner

Generalized linear models with elastic net regularization. Calls glmnet::glmnet() from package glmnet.

The default for hyperparameter family is set to "gaussian".

Details

Caution: This learner is different to learners calling glmnet::cv.glmnet() in that it does not use the internal optimization of parameter lambda. Instead, lambda needs to be tuned by the user (e.g., via mlr3tuning). When lambda is tuned, the glmnet will be trained for each tuning iteration. While fitting the whole path of lambdas would be more efficient, as is done by default in glmnet::glmnet(), tuning/selecting the parameter at prediction time (using parameter s) is currently not supported in mlr3 (at least not in efficient manner). Tuning the s parameter is, therefore, currently discouraged.

When the data are i.i.d. and efficiency is key, we recommend using the respective auto-tuning counterparts in mlr_learners_classif.cv_glmnet() or mlr_learners_regr.cv_glmnet(). However, in some situations this is not applicable, usually when data are imbalanced or not i.i.d. (longitudinal, time-series) and tuning requires custom resampling strategies (blocked design, stratification).

Dictionary

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

mlr_learners$get("regr.glmnet")
lrn("regr.glmnet")

Meta Information

Task type: “regr”
Predict Types: “response”
Feature Types: “logical”, “integer”, “numeric”
Required Packages: mlr3, mlr3learners, glmnet

Parameters

Id	Type	Default	Levels	Range
alignment	character	lambda	lambda, fraction	-
alpha	numeric	1		$[0, 1]$
big	numeric	9.9e+35		$(-\infty, \infty)$
devmax	numeric	0.999		$[0, 1]$
dfmax	integer	-		$[0, \infty)$
eps	numeric	1e-06		$[0, 1]$
epsnr	numeric	1e-08		$[0, 1]$
exact	logical	FALSE	TRUE, FALSE	-
exclude	integer	-		$[1, \infty)$
exmx	numeric	250		$(-\infty, \infty)$
family	character	gaussian	gaussian, poisson	-
fdev	numeric	1e-05		$[0, 1]$
gamma	numeric	1		$(-\infty, \infty)$
grouped	logical	TRUE	TRUE, FALSE	-
intercept	logical	TRUE	TRUE, FALSE	-
keep	logical	FALSE	TRUE, FALSE	-
lambda	untyped	-		-
lambda.min.ratio	numeric	-		$[0, 1]$
lower.limits	untyped	-		-
maxit	integer	100000		$[1, \infty)$
mnlam	integer	5		$[1, \infty)$
mxit	integer	100		$[1, \infty)$
mxitnr	integer	25		$[1, \infty)$
use_pred_offset	logical	TRUE	TRUE, FALSE	-
nlambda	integer	100		$[1, \infty)$
parallel	logical	FALSE	TRUE, FALSE	-
penalty.factor	untyped	-		-
pmax	integer	-		$[0, \infty)$
pmin	numeric	1e-09		$[0, 1]$
prec	numeric	1e-10		$(-\infty, \infty)$
relax	logical	FALSE	TRUE, FALSE	-
s	numeric	0.01		$[0, \infty)$
standardize	logical	TRUE	TRUE, FALSE	-
standardize.response	logical	FALSE	TRUE, FALSE	-
thresh	numeric	1e-07		$[0, \infty)$
trace.it	integer	0		$[0, 1]$
type.gaussian	character	-	covariance, naive	-
type.logistic	character	-	Newton, modified.Newton	-
type.multinomial	character	-	ungrouped, grouped	-
upper.limits	untyped	-		-

Offset

If a Task contains a column with the offset role, it is automatically incorporated during training via the offset argument in glmnet::glmnet(). During prediction, the offset column from the test set is used only if use_pred_offset = TRUE (default), passed via the newoffset argument in glmnet::predict.glmnet(). Otherwise, if the user sets use_pred_offset = FALSE, a zero offset is applied, effectively disabling the offset adjustment during prediction.

References

Friedman J, Hastie T, Tibshirani R (2010). “Regularization Paths for Generalized Linear Models via Coordinate Descent.” Journal of Statistical Software, 33(1), 1–22. doi:10.18637/jss.v033.i01 .

Super classes

mlr3::Learner -> mlr3::LearnerRegr -> LearnerRegrGlmnet

Methods

Inherited methods

Method `new()`

Creates a new instance of this R6 class.

Usage

LearnerRegrGlmnet$new()

Method `selected_features()`

Returns the set of selected features as reported by glmnet::predict.glmnet() with type set to "nonzero".

Usage

LearnerRegrGlmnet$selected_features(lambda = NULL)

Arguments

lambda: (numeric(1))
Custom lambda, defaults to the active lambda depending on parameter set.

Returns

(character()) of feature names.

Method `clone()`

The objects of this class are cloneable with this method.

Usage

LearnerRegrGlmnet$clone(deep = FALSE)

Arguments

deep: Whether to make a deep clone.

Examples

if (requireNamespace("glmnet", quietly = TRUE)) {
# Define the Learner and set parameter values
learner = lrn("regr.glmnet")
print(learner)

# 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 the model
print(learner$model)

# 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()
}
#> <LearnerRegrGlmnet:regr.glmnet>: GLM with Elastic Net Regularization
#> * Model: -
#> * Parameters: family=gaussian, use_pred_offset=TRUE
#> * Packages: mlr3, mlr3learners, glmnet
#> * Predict Types:  [response]
#> * Feature Types: logical, integer, numeric
#> * Properties: offset, weights
#> 
#> Call:  (if (cv) glmnet::cv.glmnet else glmnet::glmnet)(x = data, y = target,      family = "gaussian") 
#> 
#>    Df  %Dev Lambda
#> 1   0  0.00 5.3800
#> 2   1 12.71 4.9020
#> 3   1 23.27 4.4670
#> 4   2 33.34 4.0700
#> 5   2 41.88 3.7080
#> 6   2 48.98 3.3790
#> 7   2 54.86 3.0790
#> 8   2 59.75 2.8050
#> 9   2 63.81 2.5560
#> 10  2 67.18 2.3290
#> 11  2 69.97 2.1220
#> 12  2 72.29 1.9330
#> 13  2 74.22 1.7620
#> 14  2 75.82 1.6050
#> 15  3 77.23 1.4630
#> 16  3 78.48 1.3330
#> 17  3 79.52 1.2140
#> 18  3 80.38 1.1060
#> 19  3 81.10 1.0080
#> 20  3 81.70 0.9185
#> 21  4 82.34 0.8369
#> 22  4 82.95 0.7626
#> 23  4 83.46 0.6948
#> 24  4 83.88 0.6331
#> 25  4 84.22 0.5769
#> 26  4 84.51 0.5256
#> 27  4 84.75 0.4789
#> 28  4 84.95 0.4364
#> 29  4 85.12 0.3976
#> 30  4 85.25 0.3623
#> 31  4 85.37 0.3301
#> 32  5 85.50 0.3008
#> 33  5 85.61 0.2741
#> 34  5 85.70 0.2497
#> 35  5 85.77 0.2275
#> 36  5 85.84 0.2073
#> 37  5 85.89 0.1889
#> 38  5 85.93 0.1721
#> 39  5 85.97 0.1568
#> 40  5 86.00 0.1429
#> 41  6 86.02 0.1302
#> 42  6 86.05 0.1186
#> 43  6 86.07 0.1081
#> 44  6 86.09 0.0985
#> 45  6 86.11 0.0897
#> 46  6 86.12 0.0818
#> 47  8 86.24 0.0745
#> 48  8 86.42 0.0679
#> 49  8 86.56 0.0619
#> 50  8 86.68 0.0564
#> 51  8 86.78 0.0514
#> 52  8 86.86 0.0468
#> 53  8 86.93 0.0426
#> 54  8 86.99 0.0389
#> 55  9 87.04 0.0354
#> 56  9 87.08 0.0322
#> 57  9 87.11 0.0294
#> 58  9 87.14 0.0268
#> 59  9 87.17 0.0244
#> 60  9 87.19 0.0222
#> 61  9 87.20 0.0203
#> 62  9 87.22 0.0185
#> 63  9 87.23 0.0168
#> 64  9 87.24 0.0153
#> 65  9 87.24 0.0140
#> 66  9 87.25 0.0127
#> 67  9 87.25 0.0116
#> 68 10 87.26 0.0106
#> 69 10 87.27 0.0096
#> 70 10 87.27 0.0088
#> 71 10 87.28 0.0080
#> 72 10 87.28 0.0073
#> 73 10 87.28 0.0066
#> 74 10 87.28 0.0060
#> 75 10 87.29 0.0055
#> 76 10 87.29 0.0050
#> Warning: Multiple lambdas have been fit. Lambda will be set to 0.01 (see parameter 's').
#> regr.mse 
#>  10.7196

Details

Dictionary

Meta Information

Parameters

Offset

References

See also

Super classes

Methods

Public methods

Method new()

Usage

Method selected_features()

Usage

Arguments

Returns

Method clone()

Usage

Arguments

Examples

Method `new()`

Method `selected_features()`

Method `clone()`