Association rule mining is a technique widely used in the field of data mining, which consists in discovering relationships and/or correlations between the attributes of a database. However, the method brings known problems among which the fact that a large number of association rules may be extracted, not all of them being relevant or interesting for the domain expert. In that context, we propose a practical, interactive and helpful guided approach to visualize, evaluate and compare the extracted rules following a step by step methodology, taking into account the interaction between the industrial domain expert and the data mining expert.