Decision Trees

As the name points out, decision trees are used for decision making, which simply means finding the right classification for a particular situation (entity). They work by coupling simple rules in a tree-like structure. For each new entity the conditions of the single top node are compared first. For example, is variable age < 20, $20 \le$ age <40, or $40 \le$ age? According to the answer the entity is referred to one of several subnodes of the tree, each testing specialized further conditions. This process of testing conditions and referring to subnodes is iterated until we reach a leaf of the tree. Because tree construction is aimed to have leafs (almost) only reached by representatives of one class each leaf can be associated with one of the classification classes. This knowledge is then used for classification.