Using MSE to split decision tree on MLR

Question

I'm trying to split my decision tree in MLR using MSE. Here's my code

library(mlr)

cl = "classif.rpart"


getParamSet(cl)

learner = makeLearner(cl = cl
                      , predict.type = "prob"
                      #, predict.type = "response"
                      , par.vals = list(split="mse")
                      , fix.factors.prediction = TRUE
)

And it gives me the error

Error in setHyperPars2.Learner(learner, insert(par.vals, args)) : 
  classif.rpart: Setting parameter split without available description object!
Did you mean one of these hyperparameters instead: minsplit cp xval
You can switch off this check by using configureMlr!

I Know how to do this onrpart. But have no ideia on MLR

Answer 1

The split parameter is passed in a list under rpart(..., parms = list(split = "mse")). Therefore it can be set within mlr like this:

library(mlr)
cl = "classif.rpart"
learner = makeLearner(cl = cl, predict.type = "prob", par.vals = list(parms = list(split="mse")), fix.factors.prediction = TRUE)
m = train(learner, iris.task)

In the result we can see that it was passed correctly

m$learner.model$call
# rpart::rpart(formula = f, data = d, parms = list(split = "mse"), xval = 0L)