Random Forest

Bootstrapping method to introduce democracy in decision making.

General steps:

1. Generate multiple trees by bootstrapping the samples.
   • choosing times with replacement from all available training cases.
2. For each node of the tree, randomly choose variables on which t base that decision at that node.
   • You only consider these nodes to do the splitting, and ignore the rest of the nodes.
3. The final classifier outputs the class that is most agreed by the individual trees → voting!