Technology

Random Forest

Random Forest is an ensemble learning method: it aggregates predictions from a multitude of decorrelated decision trees to deliver robust classification and regression models.

Random Forest, trademarked by Leo Breiman and Adele Cutler, is a powerful supervised learning algorithm. It operates as an ensemble method, constructing hundreds of decision trees during training via a process called Bagging (Bootstrap Aggregation). Each tree is built on a random subset of the training data and, crucially, considers only a random subset of features (like $\sqrt{p}$ features) at each split. This dual-randomness strategy ensures low correlation between the individual trees, effectively reducing the high variance and overfitting that plague single decision trees. The final prediction is determined by majority vote (classification) or averaging (regression), yielding highly accurate results across diverse datasets, from financial forecasting to medical diagnosis.