Methodology for Generation and Selection of Fuzzy Classifiers of Mixed-Type Data

Abstract: A description of the method for constructing fuzzy classifiers of mixed-type data, their multi-criteria assessment and selection based on optimality principles is given. The method consists of the following main sections: 1) three-stage construction of a set of fuzzy classifiers of mixed data using Grasshopper Optimiza-tion Algorithm; 2) ranking of the obtained classifiers based on three criteria: classification error, number of features, number of rules; 3) normalization of ranks; 4) formed Pareto sets of classi-fiers; 5) selection of a fuzzy classifier based on optimality prin-ciples.

Keywords: grasshopper optimiza-tion algorithm, mixed data, fuzzy classifier, clustering