An Efficient Algorithm for Inducing Fuzzy Rules from Numerical Data

This paper proposes a modified but more powerful algorithm for inducing fuzzy if-then rules from numerical data. Data mining is performed before the process of fuzzy inference in view of the presence of noise in data. Therefore, the minimum number of learning attributes can be used to induce fuzzy rules automatically from training examples. The results obtained by this modified method demonstrate that it is more efficient and effective than relevant works in aspects of time complexity and space complexity