Comparison of data clustering methods for automatic determination of granulation in a genetic fuzzy system

The paper proposes the use of clustering methods to determine the most appropriate number of fuzzy terms when building a genetic fuzzy system. At the same time, a fuzzy logic system is used to solve data classification problems and is automatically generated by a genetic algorithm. We used a genetic algorithm with the encoding of terms and classes in a binary string, while each individual encoded a rule base. To build a rule base, it is necessary to set such a parameter as the number of fuzzy terms, since it significantly affects the quality of the generated classifiers. A comparison of the most well-known algorithms such as DBSCAN, k-means and the mean shift algorithm was carried out to identify the best data clustering meth-od. Computational experiments were carried out on several data sets to evaluate the effectiveness of the selected number of fuzzy terms. According to the results, it was determined that the mean shift algorithm selects such a number of terms that allows building more accurate classifiers in comparison with two other methods involved in testing. A comparison was also made with alternative classification methods such as k nearest neighbors, support vector machines and neural networks, as a result of which the proposed method showed comparable classification quality. The developed approach to automating the determination of the number of terms makes it possible to exclude manual selection of granulation for various data, reducing the cost of creating an effective fuzzy system for the classification problem.