You can make use of Genetic Algorithms for finding fitness of fuzzy rules generated through permutations

Hi;

take a look at this :

Efficient fuzzy rule generation: A new approach using data mining ...

sci2s.ugr.es/keel/pdf/specific/congreso/scopus7.pdf

the best way is to use fuzzy clustering if you have enough data. You should prefer interpretability-oriented methids, however, because this way you can interact with domain experts in order to maintain rules intelligently. Notice that if interpretability is not a concern, then rules are not really important. In this case you could resort to more accurate methods, suchas those used in RBF networks and SVM kernels