A new classification method using soft decision-making based on an aggregation operator of fuzzy parameterized fuzzy soft matrices

Memis S., ENGİNOĞLU S., ERKAN U.

TURKISH JOURNAL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCES, vol.30, no.3, pp.871-890, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 30 Issue: 3
  • Publication Date: 2022
  • Doi Number: 10.3906/elk-2106-28
  • Journal Name: TURKISH JOURNAL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCES
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Applied Science & Technology Source, Compendex, Computer & Applied Sciences, INSPEC, TR DİZİN (ULAKBİM)
  • Page Numbers: pp.871-890
  • Keywords: Fuzzy sets, soft sets, soft decision-making (SDM), fpfs -matrices, supervised learning, data classification, SET-THEORY
  • Çanakkale Onsekiz Mart University Affiliated: Yes

Abstract

Recently, a precise and stable machine learning algorithm, i.e. eigenvalue classification method (EigenClass), has been developed by using the concept of generalised eigenvalues in contrast to common approaches, such as k-nearest neighbours, support vector machines, and decision trees. In this paper, we offer a new classification algorithm called fuzzy parameterized fuzzy soft aggregation classifier (FPFS-AC) to combine the modelling ability of soft decision-making (SDM) and classification success of generalised eigenvalues. FPFS-AC constructs a decision matrix by employing the similarity measures of fuzzy parameterized fuzzy soft matrices (fpfs-matrices) and a generalised eigenvalue-based similarity measure. Then, it applies an SDM method based on the aggregation operator of fpfs-matrices to a decision matrix and classifies the given test sample. Afterwards, we perform an experimental study using 15 UCI datasets to manifest the success of our approach and compare FPFS-AC with the well-known and state-of-the-art classifiers (kNN, SVM, fuzzy kNN, EigenClass, and BM-fuzzy kNN) in terms of accuracy, precision, recall, macro F-score, micro F-score, and running time. Moreover, we statistically analyse the experimentally obtained data. Experimental and statistical results show that FPFS-AC outperforms the state-of-the-art classifiers in all the datasets concerning the five performance metrics.