A survey on association rule mining based on evolutionary algorithms

Searching for patterns in large database is one of the major tasks in data mining. This can be achieved by using association rule mining, which usually tends to find out the relations in an exhaustive manner. In real life, many data mining tasks require optimization between multiple objectives concurrently. Consequently over the years a large number of researches have been conducted on various techniques for efficient association rule mining and among them the field of evolutionary technique is growing rapidly with its large-scale applications and exceptional result. This research includes a systematical structured review of a wide range of state-of-the-art and recent multi-objective evolutionary algorithms (MOEAs) in terms of their chromosome representation, genetic operators and initial population, which are applied to categorical, quantitative and fuzzy rule mining problems. A lucid comparative study on various MOEA-based approaches includes computational complexity and applications within the context of this research. Finally, a guideline toward future studies on MOEA approaches has been presented that incorporates a general discussion of the current state of art literatures, newly rising study area (Interactive MOEA) along with their limitations and numerous possibilities.