A framework on rough set-based partitioning attribute selection

Authors:
Tutut Herawan;Mustafa Mat Deris
Affiliations:
Faculty of Information Technology and Multimedia, Universiti Tun Hussein Onn Malaysia, Batu Pahat, Johor, Malaysia;Faculty of Information Technology and Multimedia, Universiti Tun Hussein Onn Malaysia, Batu Pahat, Johor, Malaysia
Venue:
ICIC'09 Proceedings of the Intelligent computing 5th international conference on Emerging intelligent computing technology and applications
Year:
2009

Citing 3
Cited 2

Rough Sets: Theoretical Aspects of Reasoning about Data

Rough Sets: Theoretical Aspects of Reasoning about Data
MMR: An algorithm for clustering categorical data using Rough Set Theory

Data & Knowledge Engineering
Constructive and algebraic methods of the theory of rough sets

Information Sciences: an International Journal

The Position of Rough Set in Soft Set: A Topological Approach

International Journal of Applied Metaheuristic Computing
SAR: An Algorithm for Selecting a Partition Attribute in Categorical-Valued Information System Using Soft Set Theory

International Journal of Information Retrieval Research

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we focus our discussion on the rough set-based partitioning attribute selection. Firstly, we point out that the statement of MMR technique is an extension of Mazlack's technique is unreasonable. We prove that the mean roughness of MMR technique is only the opposite of that Mazlack's TR technique. Secondly, we observe that the suggestion of MMR to achieve lower computational complexity using the roughness measurement based on relationship between an attribute ai ∈ A and the set defined as A-{ai} instead of calculating the maximum with respect to all {aj} where ai ≠ aj, 1 ≤ i, j ≤ |A| only can be applied to a special type of information system and we illustrate this with an example. Finally, we propose an alternative technique for selecting partitioning attribute using rough set theory based on dependency of attributes in an information system. We show that the proposed technique is a generalization and has lower computational complexity than that of TR and MMR.