Minimal conditional mutual information maximisation filter calling praznik::CMIM() from package praznik.

This filter supports partial scoring (see Filter).

Format

R6::R6Class inheriting from Filter.

Construction

FilterCMIM$new()
mlr_filters$get("cmim")
flt("cmim")

See also

Examples

task = mlr3::tsk("iris") filter = flt("cmim") filter$calculate(task, nfeat = 2) as.data.table(filter)
#> feature score #> 1: Petal.Width 0.98936762 #> 2: Sepal.Length 0.05076605 #> 3: Sepal.Width NA #> 4: Petal.Length NA