Random forests are a decision tree-based ensemble learning approach. Creating numerous decision trees using bootstrapped datasets of the original data and randomly picking a subset of variables at each stage of the decision tree is what random forests are all about. The model then chooses the mode for all of the decision tree’s predictions. It lowers the possibility of inaccuracy from an individual tree by relying on a “majority wins” paradigm.
For instance, if we made a third decision tree, it would predict 0. However, the projected result would be 1 if we used the mode of all four decision trees. Random woods have a lot of power.
Random forests have numerous other advantages, including high performance, the ability to represent non-linear boundaries, no requirement for cross-validation, and the ability to prioritize features.