True or false: Unlike some other predictive modeling techniques, decision tree models do not provide confidence percentages alongside their predictions.

• A. True
• B. False
• C. Not specified
• D. Sometimes

Answer: (B)

The ability to provide a confidence or probability for a prediction is inherent in how a decision tree works. When an instance ends up in a leaf, that leaf contains the distribution of training samples that reached it—how many belong to each class. The proportion for a given class in that leaf is a probability estimate for that class, which serves as the model’s confidence in its prediction. Many tools expose this as predict_proba or similar output, giving you a percentage for each class.

Keep in mind these probabilities come from the training data seen by that leaf and can be sensitive to leaf size or class imbalances, so they may be less reliable for very small leaves. Ensemble methods like Random Forests or Gradient Boosting tend to give more stable probability estimates by aggregating across many trees. For regression trees, you can similarly use the local distribution to form confidence intervals.

So the statement is false: decision trees can provide confidence percentages alongside predictions.