一种多层次分布式数据挖掘方法的改进研究

Research on an improved multi-level distributed data mining method

针对多层次分布式数据存在高维特征和类间不平衡因素的问题,提出一种基于随机决策树检索模型的数据挖掘技术。采用随机相位重组方法进行分布式数据的层次空间重构,在重构的层次空间中提取多层次分布式数据的关联维特征量,采用高阶特征压缩方法进行降维处理,实现分布式数据的自适应挖掘。仿真结果表明,采用该方法进行数据挖掘的准确性能较好、查准率较高、计算开销降低、性能优越。

Aiming at the high?dimensional feature and inter?class imbalance factor exiting in the multilevel distributed data mining method,a multi?level distributed data mining technology based on random decision tree retrieval model is proposed.The random phase recombination method is used to reconstruct the hierarchical space of the distributed data.The correlation dimension characteristic quantity of the multi?level distributed data is extracted in the reconstructed hierarchical space,and performs the dimension reduction with the high?order feature compression method to realize the adaptive mining of distributed data.The simulation results show that the method has high accuracy for data mining,high precision ratio,low computation cost,and superior performance.