Overview of the Decision Tree

A decision tree creates a hierarchical segmentation of the input data based on a series of rules applied to each observation. Each rule assigns an observation to a segment based on the value of one predictor. Rules are applied sequentially, which results in a hierarchy of segments within segments. The hierarchy is called a tree, and each segment is called a node. The original segment contains the entire data set and is called the rootnode. A node and all of its successors form a branch. The final nodes are called leaves. For each leaf, a decision is made about the response variable and applied to all observations in that leaf. The exact decision depends on the response variable.

The decision tree requires a measure response variable or category response variable and at least one predictor. A predictor can be a category or measure variable, but not an interaction term.

The decision tree enables you to manually train and prune nodes by entering interactive mode. In interactive mode, you are unable to modify the response variable, growth properties are locked, and you cannot export model score code. Certain modifications to predictors are allowed, such as converting a measure to a category. When you are in interactive mode and modify a predictor, the decision tree remains in interactive mode, but attempts to rebuild the splits and prunes using the same rules.

To enter interactive mode, you can either start making changes to the decision tree in the Tree window or you can click Use Interactive Mode on the Roles tab in the right pane. To leave interactive mode, click Use Non-Interactive Mode on the Roles tab.