Decision trees are excellent predictive models when the dependent variable is _________.

• A. Numerical
• B. Boolean
• C. Binary
• D. Categorical

Answer: (D)

Decision trees are built to assign class labels to observations by splitting the data into regions where the target category is as pure as possible. When the dependent variable is categorical, the tree becomes a classifier: each leaf predicts a category, and splits are chosen to maximize how well they separate the classes (using measures like Gini impurity or information gain). This works for any number of classes, including the two-class (binary) case, but the general, most flexible description is a categorical target. If the target were numerical, you’d use a regression tree instead, which predicts a numeric value rather than a class label. So a categorical dependent variable is the best fit for decision trees in this context.