Probabilistic coverage of linguistic IF-THEN rules
Pavel Rusnok · 2016
Fuzzy associational analysis (FAA) is useful for solving various regression tasks. Mined fuzzy associations can be used as fuzzy IF-THEN rules in a fuzzy expert system, which is then able to predict some variable that appears on the right hand side of IF-THEN rules. In this paper we investigate fuzzy IF-THEN rules reduction methods. We propose a new probabilistic data coverage measure for evaluating quality of a set of IF-THEN rules. We also propose a rule reduction procedure based on this measure and compare it theoretically and experimentally in the context of Perception-based Logical Deduction.