Discovery of Actionable Patterns in Databases: The Action Hierarchy Approach
Gediminas Adomavičius, Alexander Tuzhilin · Faculty Digital Archive (New York University Florence) · 1997
An approach to defining actionability as a measure of interestingness of patterns is proposed. This approach is based on the concept of an action hierarchy which is defined as a tree of actions with patterns and pattern templates (data mining queries) assigned to its nodes. A method for discovering actionable patterns is presented and various techniques for optimizing the discovery process are proposed.