Fix PFI filter

[jemus42]

• Had wrong man entry linking to the performance filter rather than the permutation filter
• Standardization was performed only when standardize = FALSE, i.e. !isTRUE() instead of `isTRUE()

Reprex with behavior after fix:

library(mlr3learners)
#> Loading required package: mlr3
library(mlr3filters)
library(data.table)

set.seed(123)
task <- tgen("friedman1")$generate(n = 500)
learner <- lrn("regr.ranger", num.trees = 100)
resampling <- rsmp("holdout")
measure <- msr("regr.mse")

With standardize = FALSE

filter_pfi <- flt("permutation", 
                  learner = learner,
                  measure = measure,
                  resampling = resampling,
                  nmc = 1, 
                  standardize = FALSE)

filter_pfi$param_set$values
#> $standardize
#> [1] FALSE
#> 
#> $nmc
#> [1] 1
set.seed(123)
filter_pfi$calculate(task = task)
filter_pfi
#> <FilterPermutation:permutation>: Permutation Score
#> Task Types: regr
#> Properties: missings
#> Task Properties: -
#> Packages: mlr3, mlr3learners, ranger, mlr3measures
#> Feature types: logical, integer, numeric, character, factor, ordered
#>          feature         score
#>  1:   important2  6.5207463936
#>  2:   important4  6.4689073720
#>  3:   important1  3.9712012063
#>  4: unimportant2  0.9704977771
#>  5:   important3  0.4698455885
#>  6:   important5  0.3825112961
#>  7: unimportant3 -0.0008537673
#>  8: unimportant5 -0.7238706536
#>  9: unimportant1 -0.9496679245
#> 10: unimportant4 -1.1555031201

With standardize = TRUE

filter_pfi <- flt("permutation", 
                  learner = learner,
                  measure = measure,
                  resampling = resampling,
                  nmc = 1, 
                  standardize = TRUE)

filter_pfi$param_set$values
#> $standardize
#> [1] TRUE
#> 
#> $nmc
#> [1] 1
set.seed(123)
filter_pfi$calculate(task = task)
filter_pfi
#> <FilterPermutation:permutation>: Permutation Score
#> Task Types: regr
#> Properties: missings
#> Task Properties: -
#> Packages: mlr3, mlr3learners, ranger, mlr3measures
#> Feature types: logical, integer, numeric, character, factor, ordered
#>          feature         score
#>  1:   important2  1.0000000000
#>  2:   important4  0.9920501399
#>  3:   important1  0.6090102216
#>  4: unimportant2  0.1488323144
#>  5:   important3  0.0720539583
#>  6:   important5  0.0586606614
#>  7: unimportant3 -0.0001309309
#>  8: unimportant5 -0.1110103982
#>  9: unimportant1 -0.1456379174
#> 10: unimportant4 -0.1772041190

^{Created on 2025-08-08 with reprex v2.1.1}

[be-marc]

Thanks!