Decision trees are not only easier to understand than other predictive modeling techniques but also easier to visualize and make actionable. Another great advantage of decision tree models is their outstanding generalization when combined with other decision trees into ensembles. That is, the level of accuracy achieved by ensembles of decision trees on previously unseen data outperforms most other machine-learned models.
Moreover, ensembles of decision tree models can be applied to perform multitude of tasks such classification, regression, density estimation, manifold learning, and semi-supervised classification with thousands of real-world applications. This is a great book about their power.