A relative tolerance relation of rough set with reduct and core approach, and application to incomplete information systems

Saedudin, Rd. Rohmat (2020) A relative tolerance relation of rough set with reduct and core approach, and application to incomplete information systems. Doctoral thesis, Universiti Tun Hussein Malaysia.

Preview	Text 24p RD. ROHMAT SAEDUDIN.pdf Download (650kB) | Preview
	Text (Copyright Declaration) RD. ROHMAT SAEDUDIN COPYRIGHT DECLARATION.pdf Restricted to Repository staff only Download (334kB) | Request a copy
	Text (Full Text) RD. ROHMAT SAEDUDIN WATERMARK.pdf Restricted to Registered users only Download (25MB) | Request a copy

Abstract

Data mining concepts and methods can be applied in various fields. Many methods have been proposed and one of those methods is the classical 'rough set theory' which is used to analyze the complete data. However, the Rough Set classical theory cannot overcome the incomplete data. The simplest method for operating an incomplete data is removing unknown objects. Besides, the continuation of Rough Set theory is called tolerance relation which is less convincing decision in terms of approximation. As a result, a similarity relation is proposed to improve the results obtained through a tolerance relation technique. However, when applying the similarity relation, little information will be lost. Therefore, a limited tolerance relation has been introduced. However, little information will also be lost as limited tolerance relation does not take into account the accuracy of the similarity between the two objects. Hence, this study proposed a new method called Relative Tolerance Relation of Rough Set with Reduct and Core (RTRS) which is based on limited tolerance relation that takes into account relative similarity precision between two objects. Several incomplete datasets have been used for data classification and comparison of our approach with existing baseline approaches, such as the Tolerance Relation, Limited Tolerance Relation, and NonSymmetric Similarity Relations approaches are made based on two different scenarios. In the first scenario, the datasets are given the same weighting for all attributes. In the second scenario, each attribute is given a different weighting. Once the classification process is complete, the proposed approach will eliminate redundant attributes to develop an efficient reduce set and formulate the basic attribute specified in the incomplete information system. Several datasets have been tested and the rules generated from the proposes approach give better accuracy. Generally, the findings show that the RTRS method is better compared to the other methods as discussed in this study.

Item Type:	Thesis (Doctoral)
Subjects:	T Technology > T Technology (General) Q Science > QA Mathematics > QA71-90 Instruments and machines
Divisions:	Faculty of Computer Science and Information Technology > Department of Software Engineering
Depositing User:	Mrs. Sabarina Che Mat
Date Deposited:	03 Feb 2022 03:16
Last Modified:	03 Feb 2022 03:16
URI:	http://eprints.uthm.edu.my/id/eprint/4936

Actions (login required)

View Item