Decision trees

Tree-like representations are readable, easy to use and understand. The root of the tree is an entry node, under any node there are some branching links. The selection of a link is carried out with respect to a conditional statement assigned to the node. Evaluation of this condition determines the selection of the link. The tree is traversed top-down, and at the leaves final decisions are defined.

The example is given in Fig 2.2. Circles represent actions, rectangles hold attributes and parallelograms express relations and values.

Figure 2.2: An example of a decision tree

Decision trees could be more sophisticated. The presented decision tree (Fig 2.2) is the binary tree, every node has only two links which express two different values of a certain attribute. There are also decision trees called $\psi$-trees; in such trees a single node may have more then two links which makes a decision process more real, and allows to compare attributes with many different values in a single node. The structure of such trees is modular and hierarchical.