Tree-based Construction of Low-cost Action Rules

A rule is actionable, if a user can do an action to his/her advantage based on that rule. Actionability can be expressed in terms of attributes that are present in a database. Action rules are constructed from certain pairs of classification rules, previously extracted from the same database, each defining a preferable decision class. It is assumed that attributes are divided into two groups: stable and flexible. Flexible attributes provide a tool for making hints to a user to what changes within some values of flexible attributes are needed to re-classify a group of objects, supporting the action rule, from one decision class to another, more desirable, one. Changes of values of some flexible attributes can be more expensive than changes of other values. To investigate such cases, the notion of a cost is introduced and it is assigned by an expert to each such a change. Action rules construction involves both flexible and stable attributes listed in certain pairs of classification rules. The values of stable attributes are used to create action forest. We propose a new strategy which combines the action forest algorithm of extracting action rules and a heuristic strategy for generating reclassification rules of the lowest cost. This new strategy presents an enhancement to both methods.