Search code examples
javarstat

Using meta learning in RWeka

I used RWeka to call Weka functions directly in R.

I tried using meta learning (bagging) but failed.

My code is Bagging(classLabel ~ ., data = train, control = Weka_control(W = J48)) However, the following error pops up:

Error in Bagging(classLabel ~ ., data = train, control = Weka_control(W = J48)) : 
  unused argument(s) (data = train, control = Weka_control(W = J48))

I also tried several different base learners but always met such error.

If you successfully used meta learning in RWeka before, please let me know.

Solution

  • Just tried another writing:

    optns <- Weka_control(W = "weka.classifiers.trees.REPTree") Bagging <- make_Weka_classifier("weka/classifiers/meta/Bagging") model <- Bagging(classLabel ~ ., data=dat, control = optns)

    Surprisingly the R code works now. -Credit Leo5188