One of the important data mining techniques is association rule finding. Apriori is the most famous algorithm based on this technique. But it has a major weakness which cannot calculate the minimal value of support and confidence and these parameters is estimating intuitively by the user and this has an important effect on the algorithm performance. Main goal of this paper is to presenting an optimal method to find suitable values of minimum threshold for support and confidence by means of Binary Particle Swarm Optimization. Data used for the paper is a 4000 random records sample from Foodmart 2000 Database. Implementation of the proposed method has been done using R2010b version of MATLAB software. Proposed algorithm improves the performance of association rule mining by automatically setting suitable values for minimum support and confidence thresholds.
