CloudForest is a go/golang implementation of some utilities for analizng and scaling ensembels of decision trees developed at the Institute for Systems Biology and realesed under a modified BSD license.

It does note yet include functionality for growing ensembels of trees but will parse ensembels generated by rf-ace (http://code.google.com/p/rf-ace/ which implements Brieman and Cutler's "Random Forest" and other methods). 

Output graphs may be layed out/visualized with other ISB projects inluding:
https://github.com/ryanbressler/GraphSpectrometer
https://github.com/rbkreisberg/QED

Current Capabilities:
Parse an a predictor forest from rf-ace.
Parse an AFM feature matrix.
Apply a predictor forest to a feature matrix.
Call a closure/function with the signature of CloudForest.Recursable at each node of each tree while doing the above.

Comand Line Utilities:
leafcount
Count case-case (leaves) and case-feature (branches) coocurance and output in tsv (the only functionality availible on the comand line)
Usage of ./leafcount:
  -branches="branches.tsv": a case by feature sparse matrix of leaf cooccurance in tsv format
  -fm="featurematrix.afm": AFM formated feature matrix to use.
  -leaves="leaves.tsv": a case by case sparse matrix of leaf cooccurance in tsv format
  -rfpred="rface.sf": A predictor forest as outputed by rf-ace


Road Map:
Abstraction of collection data structures to support in memory or from disc/over network computation. 
Iterative/online computation on forests to support monitoring/steering of in progress computation.
Growing and application of Random Forests predictors in pure go (for use on app engine etc).