MATLAB Treebagger and Random Forests

http://stackoverflow.com/questions/17549337/matlab-treebagger-and-random-forests

Does the Treebagger class in MATLAB apply Breiman's Random Forest algorithm?

If I simply use Treebagger, is it the same as using Random Forests, or do I need to modify some parameters?

Thanks.

 

 

TreeBagger implements a bagged decision tree algorithm, rather than Random Forests specifically.

You can get TreeBagger to behave basically the same as Random Forests as long as the NVarsToSample parameter is set appropriately. See the documentation page for TreeBagger, under the NVarsToSample parameter, for details.

Edit: Note that in release R2015b, the NVarsToSample parameter has been renamed to NumPredictorsToSample.

 

http://stats.stackexchange.com/questions/184589/random-forests-for-predictor-importance-matlab

What you describe would be one approach. For classification, TreeBagger by default randomly selects sqrt(p) predictors for each decision split (setting recommended by Breiman).