You can get sample trees from random forest (e.g., see: How to actually plot a sample tree from randomForest::getTree()?)

However, rather than getting a single tree from random forest, is there any way to get an "optimum" or "recommended" tree which is generated considering the many trees?

Basically, I would like to replace decision tree models with random forest but would like to know what would be the best splits.

  • 1
    $\begingroup$ Related question: stats.stackexchange.com/questions/72266/… $\endgroup$ – Simone Dec 3 '13 at 23:33
  • 1
    $\begingroup$ Closing as it is effectively point 2. of that other question and got no answers for quite a long time. $\endgroup$ – user88 Dec 5 '13 at 14:05

Browse other questions tagged or ask your own question.