Machine Learning with R Cookbook - Second Edition

How it works...

The purpose of random forest is to ensemble weak learners (for example, a single decision tree) into a strong learner. The process of developing a random forest is very similar to the bagging method, assuming that we have a training set containing N samples with M features. The process first performs bootstrap sampling, which samples N cases at random, with the replacement as the training dataset of each single decision tree. Next, in each node, the process first randomly selects m variables (where m << M), and then finds the predictor variable that provides the best split among m variables. Next, the process grows the full tree without pruning. In the end, we can obtain the predicted result of an example from each single ...