task = mlr3::tsk("iris")
filter = flt("kruskal_test")
filter$calculate(task)
as.data.table(filter)
# transform to p-value
10^(-filter$scores)
if (mlr3misc::require_namespaces(c("mlr3pipelines", "rpart"), quietly = TRUE)) {
library("mlr3pipelines")
task = mlr3::tsk("spam")
# Note: `filter.frac` is selected randomly and should be tuned.
graph = po("filter", filter = flt("kruskal_test"), filter.frac = 0.5) %>>%
po("learner", mlr3::lrn("classif.rpart"))
graph$train(task)
}
Run the code above in your browser using DataLab