Absolutely. I don't think that the use of CART is sine qua non because the whole point of random forests is that it is robust and should likely be robust relative to choice of tree building method as long is the method isn't a total loss.
As such, it is likely to be be somewhat of a religious choice so providing some flexibility will be an excellent way to defuse any strongly worded discussions on the topic. Potato, Potahto as it were. On Sun, Jun 7, 2009 at 2:24 AM, Deneche A. Hakim (JIRA) <j...@apache.org>wrote: > I think that the best thing to do is to create an abstract > DecisionTreeBuilder class, this way we can use whatever implementation we > want -- Ted Dunning, CTO DeepDyve