Feature Selection via Random Search

FSelectorRandomSearch class that implements a simple Random Search.

In order to support general termination criteria and parallelization, we evaluate feature sets in a batch-fashion of size batch_size. Larger batches mean we can parallelize more, smaller batches imply a more fine-grained checking of termination criteria.

Source

Bergstra J, Bengio Y (2012). “Random Search for Hyper-Parameter Optimization.” Journal of Machine Learning Research, 13(10), 281--305. https://jmlr.csail.mit.edu/papers/v13/bergstra12a.html.

Dictionary

This FSelector can be instantiated via the dictionary mlr_fselectors or with the associated sugar function fs():

mlr_fselectors$get("random_search")
fs("random_search")

Parameters

max_features

integer(1)
Maximum number of features. By default, number of features in mlr3::Task.

batch_size

integer(1)
Maximum number of feature sets to try in a batch.

Super class

mlr3fselect::FSelector -> FSelectorRandomSearch

Methods

Public methods

• FSelectorRandomSearch$new()

• FSelectorRandomSearch$clone()

Inherited methods

• mlr3fselect::FSelector$format()

• mlr3fselect::FSelector$optimize()

• mlr3fselect::FSelector$print()

---

Method new()

Creates a new instance of this R6 class.

Usage

FSelectorRandomSearch$new()

---

Method clone()

The objects of this class are cloneable with this method.

Usage

FSelectorRandomSearch$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

library(mlr3)

terminator = trm("evals", n_evals = 5)

instance = FSelectInstanceSingleCrit$new(
  task = tsk("iris"),
  learner = lrn("classif.rpart"),
  resampling = rsmp("holdout"),
  measure = msr("classif.ce"),
  terminator = terminator
)

fselector = fs("random_search")
# \donttest{
# Modifies the instance by reference
fselector$optimize(instance)
#>    Petal.Length Petal.Width Sepal.Length Sepal.Width                 features
#> 1:         TRUE        TRUE        FALSE       FALSE Petal.Length,Petal.Width
#>     x_domain classif.ce
#> 1: <list[4]>       0.04

# Returns best scoring evaluation
instance$result
#>    Petal.Length Petal.Width Sepal.Length Sepal.Width                 features
#> 1:         TRUE        TRUE        FALSE       FALSE Petal.Length,Petal.Width
#>     x_domain classif.ce
#> 1: <list[4]>       0.04

# Allows access of data.table of full path of all evaluations
instance$archive$data()# }
#>     Petal.Length Petal.Width Sepal.Length Sepal.Width classif.ce
#>  1:        FALSE       FALSE        FALSE        TRUE       0.46
#>  2:        FALSE       FALSE        FALSE        TRUE       0.46
#>  3:         TRUE       FALSE        FALSE       FALSE       0.08
#>  4:         TRUE        TRUE        FALSE       FALSE       0.04
#>  5:         TRUE       FALSE        FALSE        TRUE       0.08
#>  6:        FALSE        TRUE         TRUE       FALSE       0.04
#>  7:         TRUE        TRUE         TRUE        TRUE       0.04
#>  8:         TRUE        TRUE         TRUE        TRUE       0.04
#>  9:        FALSE        TRUE        FALSE        TRUE       0.04
#> 10:         TRUE        TRUE         TRUE        TRUE       0.04
#>                                    uhash  x_domain           timestamp batch_nr
#>  1: e38c191b-1723-4c95-90ef-1ec22780b6ca <list[4]> 2020-10-31 04:26:25        1
#>  2: cd1320c9-1e5f-42a2-86a8-fcd20774f1e1 <list[4]> 2020-10-31 04:26:25        1
#>  3: ae06d9e1-7610-4caa-a4b7-9bf683fc569d <list[4]> 2020-10-31 04:26:25        1
#>  4: 4fff6bcd-125d-4ba8-a8fb-1a63039d03e0 <list[4]> 2020-10-31 04:26:25        1
#>  5: 01f662bd-c97d-45a2-8696-9f6e23ae83c8 <list[4]> 2020-10-31 04:26:25        1
#>  6: f42842d6-7f7a-47b3-b388-8e195c3569b2 <list[4]> 2020-10-31 04:26:25        1
#>  7: 6cfe268a-4e18-44cf-820b-23bc67816b5e <list[4]> 2020-10-31 04:26:25        1
#>  8: 897cecc3-a980-4251-b56c-d7c37ed7a2c6 <list[4]> 2020-10-31 04:26:25        1
#>  9: 48a27715-6770-4ad4-8a34-4358c2b20c1e <list[4]> 2020-10-31 04:26:25        1
#> 10: 8406e219-2663-415d-9433-295cbec8f520 <list[4]> 2020-10-31 04:26:25        1