Decision trees are suitable for classiication problems in which instances are represented by attribute-value pairs and the target concept has discrete values. A decision tree classiies an example by propagating it along a path from the root node down to a leaf node which contains the classiication for this example. Each node tests a particular attribute, each of its branches corresponds to a particular value of this attribute. Figure 9.1 shows a typical decision tree that predicts if a customer qualiies for credit based on its annual income and employment status. Each instance is classiied according to the attributes income=flow, medium,highg and employment=fnone, part time, fullg. The example employment=full, income=high, is sorted down the two right branches and categorized as a positive example according to the leaf node. Notice, that a path to a leaf node, does not necessarily use very attribute to obtain a classiication. All customers with no employment are classiied as negative examples, regardless of their income.