
Feature Selection with Sequential Search
Source:R/FSelectorSequential.R
mlr_fselectors_sequential.Rd
Feature selection using Sequential Search Algorithm.
Details
Sequential forward selection (strategy = fsf
) extends the feature set in each iteration with the feature that increases the model's performance the most.
Sequential backward selection (strategy = fsb
) follows the same idea but starts with all features and removes features from the set.
The feature selection terminates itself when min_features
or max_features
is reached.
It is not necessary to set a termination criterion.
Dictionary
This FSelector can be instantiated with the associated sugar function fs()
:
fs("sequential")
Control Parameters
min_features
integer(1)
Minimum number of features. By default, 1.max_features
integer(1)
Maximum number of features. By default, number of features in mlr3::Task.strategy
character(1)
Search methodsfs
(forward search) orsbs
(backward search).
Super class
mlr3fselect::FSelector
-> FSelectorSequential
Methods
Method optimization_path()
Returns the optimization path.
Arguments
inst
(FSelectInstanceSingleCrit)
Instance optimized with FSelectorSequential.include_uhash
(
logical(1)
)
Includeuhash
column?
Examples
# Feature Selection
# \donttest{
# retrieve task and load learner
task = tsk("penguins")
learner = lrn("classif.rpart")
# run feature selection on the Palmer Penguins data set
instance = fselect(
fselector = fs("sequential"),
task = task,
learner = learner,
resampling = rsmp("holdout"),
measure = msr("classif.ce"),
term_evals = 10
)
# best performing feature set
instance$result
#> bill_depth bill_length body_mass flipper_length island sex year
#> 1: FALSE TRUE FALSE TRUE FALSE FALSE FALSE
#> features classif.ce
#> 1: bill_length,flipper_length 0.07826087
# all evaluated feature sets
as.data.table(instance$archive)
#> bill_depth bill_length body_mass flipper_length island sex year
#> 1: TRUE FALSE FALSE FALSE FALSE FALSE FALSE
#> 2: FALSE TRUE FALSE FALSE FALSE FALSE FALSE
#> 3: FALSE FALSE TRUE FALSE FALSE FALSE FALSE
#> 4: FALSE FALSE FALSE TRUE FALSE FALSE FALSE
#> 5: FALSE FALSE FALSE FALSE TRUE FALSE FALSE
#> 6: FALSE FALSE FALSE FALSE FALSE TRUE FALSE
#> 7: FALSE FALSE FALSE FALSE FALSE FALSE TRUE
#> 8: TRUE FALSE FALSE TRUE FALSE FALSE FALSE
#> 9: FALSE TRUE FALSE TRUE FALSE FALSE FALSE
#> 10: FALSE FALSE TRUE TRUE FALSE FALSE FALSE
#> 11: FALSE FALSE FALSE TRUE TRUE FALSE FALSE
#> 12: FALSE FALSE FALSE TRUE FALSE TRUE FALSE
#> 13: FALSE FALSE FALSE TRUE FALSE FALSE TRUE
#> classif.ce runtime_learners timestamp batch_nr warnings errors
#> 1: 0.26086957 0.008 2023-03-02 12:42:40 1 0 0
#> 2: 0.20869565 0.007 2023-03-02 12:42:40 1 0 0
#> 3: 0.32173913 0.031 2023-03-02 12:42:40 1 0 0
#> 4: 0.16521739 0.007 2023-03-02 12:42:40 1 0 0
#> 5: 0.32173913 0.006 2023-03-02 12:42:40 1 0 0
#> 6: 0.49565217 0.007 2023-03-02 12:42:40 1 0 0
#> 7: 0.49565217 0.006 2023-03-02 12:42:40 1 0 0
#> 8: 0.20869565 0.008 2023-03-02 12:42:40 2 0 0
#> 9: 0.07826087 0.011 2023-03-02 12:42:40 2 0 0
#> 10: 0.15652174 0.008 2023-03-02 12:42:40 2 0 0
#> 11: 0.15652174 0.007 2023-03-02 12:42:40 2 0 0
#> 12: 0.17391304 0.009 2023-03-02 12:42:40 2 0 0
#> 13: 0.23478261 0.008 2023-03-02 12:42:40 2 0 0
#> features resample_result
#> 1: bill_depth <ResampleResult[21]>
#> 2: bill_length <ResampleResult[21]>
#> 3: body_mass <ResampleResult[21]>
#> 4: flipper_length <ResampleResult[21]>
#> 5: island <ResampleResult[21]>
#> 6: sex <ResampleResult[21]>
#> 7: year <ResampleResult[21]>
#> 8: bill_depth,flipper_length <ResampleResult[21]>
#> 9: bill_length,flipper_length <ResampleResult[21]>
#> 10: body_mass,flipper_length <ResampleResult[21]>
#> 11: flipper_length,island <ResampleResult[21]>
#> 12: flipper_length,sex <ResampleResult[21]>
#> 13: flipper_length,year <ResampleResult[21]>
# subset the task and fit the final model
task$select(instance$result_feature_set)
learner$train(task)
# }