Association rule mining is one of the most popular data mining models. Minimum support is used in the association of mining algorithms rules, such as Apriori, FP-Growth, Eclat and etc. An algorithm problem Apriori and other algorithms in the mining of association rules field, this is the user must determine the threshold of support. Assuming that the user wants to apply the Apriori algorithm to a database with millions of transactions, the user definitely can not have the necessary knowledge about all the database transactions and, therefore, could not determine an appropriate threshold. In this work, using averaging techniques, we propose a method in which the Apriori algorithm would specify the minimum support in a fully automated way. Our goal in this work is to improve the Apriori algorithm. In order to achieve this, we will initially try to use fuzzy logic to distribute the data in different clusters, and then try to introduce the user to the most appropriate threshold automatically. The results show that this approach makes any rule that can be interesting is not lost and also any rule that is useless can not be extracted.