Research On Visual Analysis Mechanism Of High-dimensional Data Based On Information Entropy

Correlation analysis of data dimensions has always been the focus of research in the field of data analysis.With the rapid development of big data in all aspects of today's society,the amount of data information has increased exponentially,and the dimension of data is getting higher and higher.The dimension disaster has become an urgent problem to be solved.Data visualization visually presents data in a graphical way,which can help users better analyze data,and visually judge the correlation between several data dimensions through graphical description.But traditional visualization methods cannot solve the problem of dimensionality disaster.Although some data mining methods are feasible,it is difficult to visualize the process,and visualization methods are still needed to provide parameter guidance in some application scenarios.In recent research,the main strategies to enhance data visualization are data reduction,visual presentation and coordination of multi view.In practical applications,composite schemes are often formed based on these three categories to help users explore high-dimensional data.In order to realize the research of the high dimensional data visual analysis mechanism based on information entropy,this paper proposes and implements the high dimensional data visual analysis prototype system ASExplorer based on information entropy.It uses a dimension importance evaluation algorithm based on joint entropy to simplify data,and designs an interactive exploration method centering on sampling scale.In order to verify the effectiveness of the prototype system,this paper verifies it from two perspectives: case study and user study.1.By analyzing the characteristics of traditional data analysis methods and combining with information entropy,a dimension importance evaluation algorithm based on joint entropy is proposed.This algorithm can help users to simplify data and select data dimension as analysis path without prior knowledge.2.Interaction technology is a method of communication between users and information systems.This paper proposes an interactive exploration method based on sampling scale.This method is based on the coordination and association multi view strategy,takes the sampling scale on the analysis path as the independent variable,calculates the change of each data dimension,and draws the change into a mini graph.The corresponding mini-graphs for each data dimension can be sorted by curve similarity.This method can simultaneously explore the association relationship of multiple data dimensions when the sampling scale of the analysis path changes,and keep the original characteristics of the data unchanged during the exploration process.3.Based on the above two points,a visual analysis prototype system named as ASExplorer is developed,which optimizes the usability of the system.Users can achieve multi-dimensional comparative analysis and hypothesis verification through simple and intuitive interactive operations such as swipe and sort.The system is suitable for the early analysis of data sets which lack prior knowledge.The effectiveness of the system is verified by case analysis and user research.