A new classification method based on rough sets theory


ÇEKİK R., Telceken S.

SOFT COMPUTING, vol.22, no.6, pp.1881-1889, 2018 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 22 Issue: 6
  • Publication Date: 2018
  • Doi Number: 10.1007/s00500-016-2443-0
  • Journal Name: SOFT COMPUTING
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.1881-1889
  • Keywords: Rough sets theory (RST), Data mining, Classification
  • Anadolu University Affiliated: Yes

Abstract

Discovering the common attributes of an object is an important problem in classification. The rough sets theory (RST) successfully reveals the relationship between an object, its attributes and classes and helps bring a solution to the classification problem. In this study, a new classification method has been developed that uses RST and a similarity-based method to create the weight matrix scoring system. The proposed method is named feature weighted rough set classification (FWRSC) and is compared with the classification methods in WEKA for five different datasets. The experimental results show that FWRSC gives higher performance than most of the methods in WEKA. Additionally, FWRSC produces the highest performance in terms of accuracy with an overall average of 67.47% for five different datasets.