Datasets and modeling

We're going to be using two of the prior datasets, the simulated data from Chapter 4, Advanced Feature Selection in Linear Models, and the customer satisfaction data from Chapter 3, Logistic Regression. We'll start by building a classification tree on the simulated data. This will help us to understand the basic principles of tree-based methods. Then, we'll move on to random forest and boosted trees applied to the customer satisfaction data. This exercise will provide an excellent comparison to the generalized linear models from before. Finally, I want to show you an interesting feature selection method using random forest, using the simulated data. By interesting, I mean it's a valuable technique to add to your feature selection arsenal, but I'll point out a couple of caveats for you to consider in practical application.