Usage of decision trees

Decision trees belong the most accurate classifiers. Unlike other algorithms, decision trees can handle discrete attributes and numerical attributes by using them under split conditions that are represented by symbols.

Note: This feature is available starting from Db2® version 11.5.4.

Another advantage is the human readability of decision tree models. Particularly applications that require both, high accuracy and interpretability of the classification model, can use decision trees. Such applications are, for example, customer classification, fraud detection, and diagnostics.