数据可视化研究综述
作者：刘滨 刘增杰 刘宇 李子文 陈莉 孙中贤 王莹 张一辉 赵佳盛 张红斌 刘青
来源：《河北科技大学学报》2021年第06期
        摘 要：数据可视化对于从海量数据中发现规律、增强数据表现、提升交互效率具有重要作用。目前，数据可视化的概念及相关研究领域不断扩展，就数据类型而言，可视化研究逐渐聚焦于多维数据、时序数据、网络数据和层次化数据等领域。通过对中国知网（CNKI）中外文文献进行分析可知：2014年、2015年是数据可视化领域研究热度升级、理论成果大量产出的“里程碑”式年份;中国大数据领域研究热潮形成后，数据可视化是迅速发展的一个重要支撑领域;国内外数据可视化领域的研究，在时间上基本同步，而武汉大学、浙江大学、北京邮电大学、国防科技大学、电子科技大学等都是在该领域研究活跃度较高的国内高校。
        要获得良好的视觉效果，帮助用户降低理解难度，高效分析数据和洞悉价值，通常还需要注意彩与语义、突出核心数据、防止数据过载、防止思维过度发散等技术要点。现有的
数据可视化技术主要分为基于几何技术、基于图标技术、基于降维技术、面向像素技术、基于时间序列技术、基于网络数据技术的数据可视化方法，以及层次可视化技术和分布技术等。基于几何技术的可视化方法，包括平行坐标、散点图矩阵、Andrews曲线等。基于坐标的可视化方法，可以清晰展示变量间的关系，但受限于屏幕尺寸，当数据维度超过3个时，难以直观显示全部维度，需要结合人机交互技术进行展示，适用于表达不同维度之间的相关关系，比如学生学习行为之间的关联关系等。基于图标的可视化方法，主要包括星绘法和Chernoff面法，以几何图形作为图标刻画多维数据，直观反映出图标各个维度所表示的意义，适用于工作完成情况、激励工作进度概览等。基于降维技术的可视化方法，根据维度属性确定点的坐标，在保持数据关系不变的前提下映射到低维可视空间中，主要涉及主成分分析、自组织映射、等距映射等。基于时间序列的可视化方法，是一种显示数据间相互关系和影响程度的可视化方法，主要包含线形图、堆积图、地平线图等，随着时间发展采集相应数据，并利用上述3类可视化方法进行呈现，适用于表示信息数据流动和变化状态，如不同时间段成绩流向趋势分布、主题概念的变迁等。基于网络数据的可视化方法，核心是自动布局算法，通过自动布局与计算绘制成网状结构图形，主要有力导向布局、圆形布局、网格布局等，常用来表示大规模社交网络结构，适用于活跃度分析、引文关系展现等。层次可视化技
术，主要包括节点链接、空间填充、混合方法等，通过绘制不同形状的节点和包围框来表示层次结构的数据，适用于表示组成员间交互关系的发现和挖掘，如在线协作员工之间的交互。
        基于CNKI，通过对数据可视化研究情况的分析，提出数据可视化研究过程中的注意点，指出数据可视化需要重点考虑彩的匹配，在彩与数据内容的重要度之间建立关联;可视化方案应在满足业务需求的基础上以业务逻辑为依据，合理组合与应用相关可视化技术;统一的可视化风格有助于提升人们理解数据的连贯性、一致性和效率，兼顾用户的审美要求，在风格与彩之间建立合理的匹配关系;数据可视化应以实用、合理、高效地表现关键过程、关键目标、关键结果为主要面向。此外，对可视化应用实例Echarts展开综述，包括Echarts 交互组件（markPoint和markLine标注点组件、dataZoom区域组件、图例交互组件）在可视化中的应用，以及动态数据绘制等。最后，对可视化存在的挑战以及未来研究方向进行了分析和展望，指出虚拟现实、可视化系统和数据分析是可视化未来的研究方向，其应用热点领域还包括统计可视化、新闻可视化、思维可视化、社交网络可视化和搜索日志可视化等。
        关键词：计算机图形学;数据可视化;多维数据;时序数据;网络数据;层次化数据
        中图分类号：TP393 文献标识码：A
        doi：10.7535/hbkd.2021yx06012
        Review of data visualization research
        LIU Bin1，2，LIU Zengjie1，2，LIU Yu3 ，LI Ziwen4，CHEN Li5，SUN Zhongxian1，2，WANG Ying1，2，ZHANG Yihui1，2，ZHAO Jiasheng1，2，ZHANG Hongbin6，LIU Qing1，2
        （1.School of Economics and Management，Hebei University of Science and Technology，Shijiazhuang，Hebei 050018，China;2.Research Center of Big Data and Social Computing，Hebei University of Science and Technology，Shijiazhuang，Hebei 050018，China;3.Library，Hebei Professional College of Political Science and Law，Shijiazhuang，Hebei 050061，China;4.Hebei Institute of Laser Company Limited，Shijiazhuang，Hebei 050081，China;5.Air Force Early Warning Academy，Wuhan，Hubei 430019，China;6.School of Information Science and Engineering，Hebei University of Science and Techno-logy，Shijiazhuang，Hebei 050018，China）
数据可视化的概念
        Abstract：
        Data visualization plays an important role in discovering rules from massive data，enhancing data performance and improving interaction efficiency.At present，the concept of data visualization and related research fields are expanding.In terms of data types，the current visualization research gradually focuses on the fields of multidimensional data，time series data，network data and hierarchical data.Through the analysis of Chinese and foreign literature on CNKI，
        it can be seen that 2014 and 2015 are "milestone" years in which the research heat in the field of data visualization is upgraded and a large number of theoretical achievements are produced;Data visualization is an important supporting field of rapid development after the formation of the research upsurge in the field of big data in China;The research in the field of data visualization at home and abroad has basically achieved synchronization in time;Wuhan University，Zhejiang University，Beijing University of Posts and telecommunications，University of national defense science and technology and University
of Electronic Science and technology research actively in this field in China.In order to obtain good visual effects，help users reduce the difficulty of understanding，efficiently analyze data and insight value，It is usually necessary to pay attention to technical points such as color and semantics，highlighting core data，preventing data overload and preventing excessive divergence of thinking.The existing data visualization technologies are mainly divided into geometry based technology，icon based technology，dimension reduction based technology，pixel oriented technology，time series based technology，network data based technology，hierarchical visualization technology and distribution technology.Visualization methods based on geometric technology，including parallel coordinates，scatter matrix，Andrews curve，etc;The coordinate based visualization method can clearly show the relationship between variables，but limited by the screen size，it is difficult to visually display all dimensions when the data dimensions exceed three.It needs to be displayed in combination with human-computer interaction technology，which is suitable for the correlation between different dimensions，such as the correlation between students' learning behaviors;Icon based visualization method mainl
y includes star drawing method and Chernoff surface method.Geometric graphics are used as icons to depict multi-dimensional data，which intuitively reflects the visual significance of each work surface.It is suitable for work completion and incentive work progress overview，etc;The visualization method based on dimension reduction technology determines the coordinates of points according to the dimension attributes and maps them to the low-dimensional visual space on the premise of keeping the data relationship unchanged.The dimension reduction technology mainly involves principal component analysis，self-organizing mapping，isometric mapping，etc;The visualization method based on time series is a visualization method to display the relationship and influence degree between data，mainly including linear graph，stacking graph，horizon graph，etc.the corresponding data is collected with the development of time and presented by the above three visualization methods，which is suitable for representing the flow and change state of information data，such as the trend distribution of grades in different time periods and the change of theme concepts，etc;The core of the visualization method based on network data is the automatic layout algorithm，which draws the graph of network structure
through automatic layout and calculation.It mainly strongly guides the layout，circular layout and grid layout，etc.It is commonly used to represent the large-scale social network structure，which is suitable for activity analysis，citation relationship，etc;Hierarchical visualization technology mainly includes node connection，space filling and hybrid methods，etc.it represents the data of hierarchical structure by drawing nodes and bounding boxes with different shapes.It is suitable for the discovery and mining of interactive relationships among group members，such as the interaction between online collaborative employees.

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。