Decision trees

Decision trees are popular since they can roughly be equated with If/Then/Else rules used in some business contexts and are relatively easy to explain to managers. Decision trees are not only used for classification. When they are used to predict numeric outcomes, they are referred to as regression trees. The basic concept that decision trees use is that each node of the tree is split into two parts based upon an optimal split point. The tree continues to grow by adding more leafs until it is not able to make any additional splits, which improve the ability to distinguish between any of the decisions.