You could always use R with RapidMiner. I'm not an R expert but I found a package called "rpart" for recursive partitioning and "earth" for multivariate adaptive regression spline models (mars - ha ha). Random forests and boosting are already operators in RapidMiner.
Learning curve can be extensive but if you know another tool, I guess you would have the jargon learned already which in my experience is a great help. My personal experience is that it took a few months to really cross the line and become productive. There are many sources of data out there that help. I even maintain a low frequency blog http://rapidminernotes.blogspot.com/
to give something back to the community where I share the occasional thought about some tips and tricks I've learned on the way.