A Formalism to Extract Fuzzy If-Then Rules from Numerical Data Using Genetic Algorithms

In many applications knowledge required has to extract from a massive amount of numerical data. In this paper, extracting fuzzy if-then rules from numerical data is discussed. Due to The comprehensibility of fuzzy if-then rules is related to various factors. Our discussion is concentrated on simplicity of fuzzy rule-based systems, i.e., optimizing the number of input variables and the number of fuzzy if-then rules. Firstly, extracting fuzzy rule from numerical data is considered in decision information system, and confidence and support of fuzzy rule are obtained. Then, by encoding fuzzy partition and membership functions, selecting weighted mean of confidence and support of fuzzy rule as fitness function, optimizing the number of if-then rule and its inputs are formally discussed based on genetic algorithms (GAs)