数据集压缩建模的研究

Research on Data Set Compression Modeling

目前金融、医疗、农业等各行各业都处在大数据的背景下,数据集的处理问题已成为研究的热点。数据集的有效处理有利于从海量的数据中挖掘出有价值的信息,通过对数据序列变化趋势的分析和预测,揭示事物的内在规律和关联。论文在分析了已有模型的基础上,提出了两种新的模型。数据集的压缩模型以及奇异值的识别模型。经实验证明,使用论文提出的压缩模型进行数据集压缩,不仅考虑到数据集的时间特性,而且与已有模型相比,出错率最高能降低16.1%。同时,在奇异值识别研究中,所提出的模型,压缩率最高能达到92%,与原数据序列的差异值最小为0.03。

At present,finance,medical care,agriculture and other industries are in the context of big data,data set processing has become a research hotspot.The effective processing of data sets is conducive to mining valuable information from massive data and revealing the internal laws and associations of things through the analysis and prediction of the trend of data sequence changes.Based on the analysis of existing models,two new models are proposed in this paper.Data set compression model and singular value recognition model.Experimental results show that the proposed compression model can not only take into account the time characteristics of the data set,but also reduce the error rate by 16.1%compared with the existing models.At the same time,in the study of singular value recognition,the compression rate of the proposed model can reach 92%,and the difference between the proposed model and the original data sequence is 0.03.