Title: An application of the logic of explanatory power in rough set analysis: implications for the classification of decision rules

Authors: Anthony T. Odoemena

Addresses: Department of International Studies, University of Tokyo, Kashiwanoha, 277-8563, Japan

Abstract: This paper uses the logic of explanatory power to address the question of uncertain decision rule classification and interpretation in rough set data analysis. A set theoretic configuration of the measure of explanatory power is introduced. The usefulness of the measure is then examined in the context of two datasets - one related to car evaluation and the other related to the provision of extra educational supports. It is found that the explanatory power measure has some interesting properties that enhance the informativeness and interpretation of non-deterministic decision rules. The result of the numerical analysis shows that the explanatory power index is unique. The index can also facilitate the establishment of an objective threshold that determines whether the explanatory relevance of the premise in a given decision rule is positive, negative, or neutral.

Keywords: rough sets; explanatory power; data analysis; decision rules.

DOI: 10.1504/IJDS.2019.100329

International Journal of Data Science, 2019 Vol.4 No.2, pp.85 - 100

Received: 09 Jan 2018
Accepted: 26 Jun 2018

Published online: 26 Jun 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article