Title: A Method For Evaluating An Action Rule Specified By A User

Year of Publication: Nov - 2014
Page Numbers: 22-27
Authors: Seunghyun Im
Conference Name: The International Conference on Data Mining, Internet Computing, and Big Data (BigData2014)
- Malaysia

Abstract:


An action rule is a rule extracted from a decision table that describes the expected change in the decision value of an object as a result of the changes made to some of its condition values. We use it primarily to recommend ways to improve the current state of an object. e.g., improve a person's health, student retention rate, or manufacturing process. Various methods have been proposed to automatically generate concise and effective action rules. However, a practical problem not discussed in the existing literature is the method for evaluating an action rule specified by a user. This is an important task because, in many cases, users (e.g. the decision maker of an organization) have some knowledge about specific actions they can take, and need to know the performance of the action rule to make a better decision. In this paper, we present a taxonomy of the types of action rules submitted by a user and a method for measuring the performance of the rules.