基于改进的BPSO算法的关联规则挖掘

Association Rules Mining Based on Improved BPSO Algorithm

为了从大数据集中挖掘关联规则,提出了一种改进的二进制粒子群优化算法(GRBPSO)用于挖掘关联规则。首先,结合关联规则设计BPSO算法的适应度函数,然后对种群进行预处理,保证初始种群的质量,最后设计一种缩减搜索空间的优化策略,以减少搜索空间。基于六个高维数据集,将GRBPSO算法与普通BPSO算法进行比较以证明其有效性;将GRBPSO算法与PSOFIM算法、GA-Apriori算法及PSO-Apriori算法进行比较以证明其可行性。实验结果表明,GRBPSO算法具有一定的可行性且挖掘效率更高。

In order to mine association rules from big data sets,an improved binary particle swarm optimization algorithm(GRBPSO)is proposed to mine association rules.Firstly,the fitness function of BPSO algorithm is designed combined with associa⁃tion rules,and then the population is preprocessed to ensure the quality of the initial population,and finally an optimization strategy is designed to reduce the search space.Based on six high-dimensional data sets,GRBPSO algorithm is compared with the common BPSO algorithm to prove its effectiveness.GRBPSO algorithm is compared with PSOFIM algorithm,GA-Apriori algorithm and PSO-Apriori algorithm to prove its feasibility.Experimental results show that GRBPSO algorithm is feasible and efficient.