Setting a rule base for a fuzzy classifier using the grasshopper optimization algorithm and the clustering algorithm

Abstract

The article presents a description of a hybrid algorithm for generating fuzzy rules for a fuzzy classifier using grasshopper optimization algorithm and the K-means data clustering algorithm. The performance of clustering was evaluated by three fitness functions: total variance, Davis–Bouldin index, and Calinski–Harabasz index. Triangular and Gaussian membership functions have been investigated. The efficiency of the generated fuzzy rule bases has been tested on real datasets. The best combination is to use the total variance as the fitness function and the Gaussian function as the membership function.