Abstract
This study proposes a novel methodology for the extraction of a hierarchical Takagi-Sugeno fuzzy rule-based architecture from data. This architecture reduces the number and complexity of involved fuzzy rules, and, in many cases, improves the predictive performance of the model. The proposed hierarchical architecture takes the form of a cascading topology in which the predicted result computed at the previous layer is considered in the output part of the fuzzy rules. We propose a well-defined general methodology for the extraction of this hierarchical topology from data and discuss strategies for feature selection and choosing the number of rules at each level. The performance of the proposed methodology is demonstrated through extensive experiments, including case studies outlining specific behaviors and parameterizations, and comparative experiments showing the performance of the proposed architecture compared to a standard flat fuzzy rule-based system. (C) 2019 Elsevier B.V. All rights reserved.