By default, Minitab displays the smallest tree with a misclassification cost that is within 1 standard error of the minimum cost.
The root node is the top node of the tree. It is the only node that does not have parents. At each node, the data is split into 2 children, depending on the characteristics of the data. For example, does a heart patient have exercise angina? Nodes continue to split until the tree runs out of data to split. In other words, the terminal nodes cannot be split into further groupings.
When you validate the performance of the tree with a test data set, the tree diagram shows cases from the training data set.
The paths to the terminal nodes in the tree diagram describe how purer groups are formed. The terminal nodes provide information on how to define each purer group with the predictor variables for further study. Each node contains details on the class, the count, and the percentage in the class.
The information contained in the terminal nodes also provides a way to rank all the purer groups by class probabilities, in order to focus your efforts.