三段式编码的改进的IGA关联规则挖掘算法

Association Rule Mining Strategy Based on Three-Sectional Coding Immune Genetic Algorithm

最新智能算法在关联规则挖掘上存在挖掘精度低,易陷入局部收敛,运行时间较长等弊端,针对以上问题,提出了求解连续属性关联规则挖掘的三段式的改进的免疫遗传挖掘算法(TIIGA),首先使用三段式编码方案降低分割点的选取对挖掘的影响,其次提出了基于矢量矩浓度的TIIGA的选择方案,可以提高挖掘规则的多样性和挖掘的精度,最后使用了自适应的交叉与变异因子降低人工设置参数对挖掘结果的干扰。实验结果表明,与最新智能算法相比,提出的TIIGA算法在关联规则连续属性挖掘上具有挖掘精度高、全局收敛,挖掘时间短等优势。

There are some shortcomings of low mining accuracy and falling into local convergence easily in the latest intelligence algorithm on the mining association rules mining. To solve these problems, a TIGA strategy was propose. Firstly, a three-step encoding was used to encode continuous association rule mining in order to educe the segmentation point of mining influence. Secondly, an immune algorithm was used to mine the association rules. A multidimensional mining plan was proposed based on vector distance of genetic algorithm, which can increase the diversity of population and the accuracy of mining rules. Finally, the adaptive crossover and mutation factors were uses to re- duce the interference of artificial setting parameters on the mining results. The experimental results show that, compared with the latest mining algorithm, the proposed algorithm has the advantages of high precision and global convergence based on mining association rules.