Summary

Random forests were created as an improvement on the bagging method. As an example of the homogeneous ensemble method, we saw how the forests help in obtaining higher accuracy. Visualization and variable importance for random forests were thoroughly detailed. We also saw a lot of diagnostic methods that can be used after fitting a random forest. The method was then compared with bagging. Novel applications of random forest for missing data imputation and cluster analysis were also demonstrated.

In the next chapter, we will look at boosting, which is a very important ensemble.