The linear models we studied in Chapters 7, Linear Models and Chapter 8, Time Series Models, learn a set of parameters to predict the outcome using a linear combination of the input variables, possibly after transformation by an S-shaped link function in the case of logistic regression.
Decision trees take a different approach: they learn and sequentially apply a set of rules that split data points into subsets and then make one prediction for each subset. The predictions are based on the outcome values for the subset of training samples that result from the application of a given sequence of rules. As we will see in more detail further, classification trees predict a probability estimated from the ...
