A decision tree is an interpretable supervised learning model that represents decision rules in a tree-like structure. Inference follows a path from a root node through feature-based conditions to a leaf node that provides the predicted class or value.