Leo Breiman and Adele Cutler have originally devised the idea at the core of the Random Forests algorithms, and the name of the algorithm remains today a trademark of theirs (though the algorithm is open source). Random Forests are implemented in scikit-learn as RandomForestClassifier / RandomForestRegressor.
Random Forests works in a similar way as bagging, also devised by Leo Breiman, but it operates only using binary split decision trees, which are left to grow to their extremes. Moreover, it samples the cases to be used in each of its models using bootstrapping. And, as the tree is grown, at each split of a branch, the set of variables to be considered for the split is drawn randomly, too. In the end, that's ...
