Desktop Survival Guide
by Graham Williams
Random forests can also be used in an unsupervised mode for clustering. See Unsupervised Learning with Random Forest Predictors at http://www.genetics.ucla.edu/labs/horvath/RFclustering/RFclustering.htm.
Random forests were introduced by (), building on the concept of bagging (, ) and the random subspace method for decision forests (, ). Breiman observed that ``random forests do not overfit.''