The Research And Implementation Of Exploratory High-dimensional Data Visualization System Based On Web

With the increasing number of available computing resources and sensor devices,the number of data is exploding.While the amount of data is increasing,the dimension of data is getting higher and higher.How to enable users to quickly understand and analyze a large number of complex high-dimensional data is a hot topic of current research.Data visualization can display the data intuitively in the form of charts,and help users to better analyze the data.At present,many researchers have studied the direction of high-dimensional data visualization.But there are still some problems:(1)The current visualization systems have poor support for the display of high-dimensional data,and usually display the massive high-dimensional data directly,resulting in data obscuring each other and creating visual confusion;(2)Most of the visualization platforms focus on improving the visualization effect,but lack of effective guidance for analysts to understand the characteristics of the data and improve the quality of the data;(3)The exploration process of visualization platform is not perfect.The platforms don't have good versatility and interactivity,and the threshold for use is high.These problems need to be solved urgently,so the visualization of high-dimensional data is studied deeply in this paper.The main contents of this paper are as follows:(1)On the basis of deep understanding of the characteristics of high-dimensional data and the visualization technology of parallel coordinates,a visual analysis method of high-dimensional data based on parallel coordinates is designed and implemented.In this method,dimension reduction,edge binding and visual interaction are used to optimize the parallel coordinates.And to solve the problem of visual confusion in the process of visualization,an edge binding algorithm based on AGNES and K-means double layer clustering is proposed.The experimental results show that the algorithm has higher accuracy than other common edge binding algorithms.(2)By analyzing and studying the characteristics of traditional data analysis methods,combined with data visualization technology,a new exploratory visual analysis method is designed and implemented.In this method,the process of exploratory visual analysis is divided into three parts:attribute type analysis,attribute feature analysis and attribute relationship analysis.In the analysis steps,with the help of rich and effective visualization means and interaction technology,to improve the understanding and analysis ability of analysts to unknown high-dimensional data.(3)The design and implementation of a sound visual analysis process and good interactivity of the system.The system includes many functional modules,such as data import,feature engineering,exploratory visual analysis,high-dimensional data visualization and visualization application,which reduces the threshold for the analysis and visualization of high-dimensional data.Based on the above research content and achievement,this paper designs and realizes the exploring high-dimensional data visualization system based on Web.The system supports a variety of data sources,and provides a perfect high-dimensional data visualization exploration process for users.The system also integrates a variety of high-dimensional data visualization layout components and analysis algorithms,with good interactivity and scalability.It can help users to explore the law and value of high-dimensional data accurately and efficiently,and improve the comprehensibility and usability of the data.