MOFOR_QAR:Multi-objective Fireworks Optimization Algorithm To Mine Positive And Negative Quantitative Association Rules

With the massive growth of data, traditional data analysis has been unable to cope with demand of people for discovering knowledge from massive data. Consequently, data mining technology emerges as the times require. Used to find interesting links between attribute sets, association rule mining is an important part of data mining research, and one of the important research topic, quantitative association rule mining, are getting more and more attention because of their wide applications in many fields, such as business.A careful study of the current domestic and foreign for quantitative association rules mining research status and related mining methods was made. Based on this, a multi-objective fireworks optimization algorithm for the mining of positive and negative quantitative association rules was launched. Firstly, the basic concepts of association rules, relevant evaluation index and related mining algorithms was introduced, then introduced the related concepts of multi-objective optimization problem and solving methods, the focus was the problems existing in the existing association rule mining algorithms. Based on the study, adequate research of fireworks optimization, a swarm Intelligence Algorithm, was conducted, then the multi-objective fireworks optimization algorithm to mine positive and negative quantitative association rules was proposed. Compared with the existing quantitative association rules mining methods, the method has the following improvements: first, both positive and negative quantitative association rules can be obtained, the problem was solved as a multi-objective optimization problem, multiple objectives, to balance multiple objectives can be optimized simultaneously, and the impact of human intervention was reduced; second, the contractile mechanism makes the gotten association rules be more accurate; third, the introduction of the redundancy elimination mechanism based similarity can effectively maintain the diversity of the association rules.In order to verify the effectiveness of the algorithm, a series of experiment based on real datasets was performed. Compared with other similar algorithms, the algorithm can obtain stable results on different data sets. the dataset can be fully covered, and a good balance between reliability, correlation and intelligibility was obtained.