In this paper, a fuzzy association rule mining approach with type-2 membership functions is proposed for dealing with data uncertainty. It first transfers quantitative values in transactions into type-2 fuzzy values. Then, according to a predefined split number of points, they are reduced to type-1 fuzzy values. At last, the fuzzy association rules are derived by using these fuzzy values. Experiments on a simulated dataset were made to show the effectiveness of the proposed approach.
Intelligent Information and Database Systems Part II LNAI 9012, pp.128-134