A classification method based on Hamming pseudo-similarity of intuitionistic fuzzy parameterized intuitionistic fuzzy soft matrices

Authors : Samet Memiş, Burak Arslan, Tuğçe Aydin, Serdar Enginoğlu, Çetin Camci

Pages : 59-76

View : 19 | Download : 7

Publication Date : 2021-08-31

Article Type : Research

Abstract :In this study, firstly, Hamming pseudo-similarity of intuitionistic fuzzy parameterized intuitionistic fuzzy soft matrices (ifpifs-matrices) have been defined. Afterwards, a classifier based on Hamming pseudo-similarity of ifpifs-matrices (IFPIFS-HC) has been developed. The classifier's simulations have been performed using datasets provided in the UCI Machine Learning Database, and its performance results via the performance metrics accuracy, precision, recall, macro F-score, and micro F-score have been obtained. Thereafter, the results have been compared with those of the well-known methods. Then, the statistical evaluations of the performance results have been conducted using Friedman and Nemenyi post-hoc tests, and the critical diagrams of the Nemenyi post-hoc test are presented. The results and the statistical evaluations show that the proposed classifier has performed better than the others in 12 of 21 datasets in terms of the five performance metrics, in 4 of 21 in terms of the four performance metrics, and 17 of 21 in terms of accuracy performance metric. Moreover, the mean accuracy, precision, recall, precision, macro F-score, and micro F-score results of Fuzzy kNN, FSSC, FussCyier, HDFSSC, and FPFS-EC for the 21 datasets are 84.90, 71.96, 67.95, 71.91, and 75.28; 78.12, 68.01, 68.05, 66.53, and 67.68; 80.76, 68.63, 69.07, 68.36, and 70.65; 81.93, 69.43, 69.95, 70.25, and 72.36; and 89.59, 80.27, 78.40, 81.20, and 83.60, while those of IFPIFS-HC are 90.59, 82.88, 80.75, 82.89, and 85.48, respectively. Finally, the applications of ifpifs-matrices to machine learning have been discussed for further research.
Keywords : Soft sets, Intuitionistic fuzzy sets, ifpifs-sets, ifpifs-matrices, Similarity measures, Machine learning