Decision tree

A decision tree is a supervised learning model that partitions data through a sequence of if-then splits on individual features. Each internal node tests a threshold; each branch leads to another split or to a leaf that assigns a class label or numeric prediction. The structure is easy to read, which made trees attractive in business and medicine long before neural networks returned to prominence.

A single deep tree can overfit noise by memorising training quirks. Ensemble methods address that weakness: bagging trains many trees on bootstrap samples and averages their votes, and Leo Breiman’s random forest adds random feature subsets at each split to further decorrelate the trees. See the random forests article and timeline entry.