Hybrid Grid Partition and Rought Set Methods for Generating Fuzzy Rules in Data Set Classification

Abstract

This research aims to address the issue of exponential rule generation in fuzzy rule-based classification systems by developing a hybrid grid partition and rough set method. Fuzzy rule-based classification systems have the potential to construct linguistically understandable models, but a major constraint is the significant increase in the number of rules with a high number of attributes, which can diminish interpretation and classification accuracy. In this study, the grid partition method is utilized to generate fuzzy rules with adaptively adjusted grid structures, thus avoiding exponential rule proliferation. The research encompasses the use of the Iris Flower dataset, rule formation while considering variable precision, and classification accuracy testing. The research findings indicate that the hybrid grid partition and rough set method produces more efficient and accurate fuzzy rules, with a classification accuracy rate of 83.33%. This method also successfully reduces the number of generated rules, making it a promising solution to tackle the issue of exponential rule increase in fuzzy rule-based classification systems