MDataEE:多因素时间序列数据的分析与可视化  

MDataEE:Analysis and Visualization of Multifactor Time Series Data

作　　者：路强 葛逸凡[2] 余烨 黎杰[1,2,3] 饶金刚 Lu Qiang;Ge Yifan;Yu Ye;Li Jie;Rao Jingang(Key Laboratory of Knowledge Engineering with Big Data(Hefei University of Technology),Hefei 230009;School of Computer and Information,Hefei University of Technology,Hefei 230601;Intelligent Interconnected Systems Laboratory of Anhui Province(Hefei University of Technology),Hefei 230009;Huangshan Tourism Group Hydropower Development Co.,LTD.,Huangshan 245000)

机构地区：[1]大数据知识工程教育部重点实验室(合肥工业大学),合肥230009 [2]合肥工业大学计算机与信息学院,合肥230601 [3]智能互联系统安徽省实验室(合肥工业大学),合肥230009 [4]黄山水务控股集团有限公司,黄山245000

出　　处：《计算机辅助设计与图形学学报》2022年第10期1613-1625,共13页Journal of Computer-Aided Design & Computer Graphics

基　　金：安徽省重点研发计划(201904D07010);黄山市科技计划(2019KN-05);合肥工业大学智能制造研究院科技成果培养(IMIPY2021022).

摘　　要：多因素时间序列数据及异常数据的可视化对于提高决策分析效率等问题具有十分重要的意义.由于不同种类数据具有不同的特征,传统的可视化方法在绘制此类数据时会出现图像复杂、用户观察效率低的情况.为此,提出一种高效探索多因素时间序列数据及异常数据的可视化方法MDataEE.首先,使用可视化映射简化多种类数据的视图;其次,根据数据的密度和重要性以及视觉感知来优化坐标轴的绘制;最后,增加了一些交互操作,通过图像显隐及生成对比视图等操作,方便用户根据需求自由探索不同方面的数据.在真实PM2.5数据集上进行的实验结果表明,与传统可视化方法相比,所提出的方法能够生成简洁且易于分析的可视化视图,在分析异常数据变化的趋势及原因等方面更有优势,可提高用户理解并分析异常的多因素时间序列数据的效率.The visualization of multifactor time series data and abnormal data is of great significance in improving the decision-making efficiency and other problems.Since different types of data have different characteristics,traditional visualization methods will face the challenges of complex graphics and low user observation efficiency when drawing such data.Therefore,an efficient visualization method MDataEE for exploring multifactor time series data and abnormal data is proposed.Firstly,the view of multiple kinds of data is simplified by visual mapping.Secondly,the rendering of coordinate axes is optimized according to the density and importance of data and visual perception.Finally,we added some interactive operations,such as displaying and hiding graphics and generating contrast views,so that users could explore different aspects of data freely according to their needs.The real-world PM_(2.5) data set is used for the experimental tests in this paper.The results show that the proposed method can generate a concise visual view,which has advantages in analyzing the trend and causes of abnormal data,and can improve the efficiency in understanding and analyzing abnormal multifactor time series data.

关 键 词：多因素时间序列数据 异常数据 可视化设计 可视化分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]