Irene Kahvazadeh and Mohammad Saniee Abadeh, Tarbiat Modares University, Iran

Extracting association rules from numeric features involves searching a very large search space. To deal with this problem, in this paper a meta-heuristic algorithm is used that we have called MOCANAR. The MOCANAR is a Pareto based multi-objective cuckoo search algorithm which extracts high quality association rules from numeric datasets. The support, confidence, interestingness and comprehensibility are the objectives that have been considered in the MOCANAR. The MOCANAR extracts rules incrementally, in which, in each run of the algorithm, a small number of high quality rules are made. In this paper, a comprehensive taxonomy of meta-heuristic algorithm have been presented. Using this taxonomy, we have decided to use a Cuckoo Search algorithm because this algorithm is one of the most matured algorithms and also, it is simple to use and easy to comprehend. In addition, until now, to our knowledge this method has not been used as a multi-objective algorithm and has not been used in the association rule mining area. To demonstrate the merit and associated benefits of the proposed methodology, the methodology has been applied to a number of datasets and high quality results in terms of the objectives were extracted.

Numeric Association Rule, Cuckoo Search, Multi-Objective Algorithm