The Research On Visual Data Mining Technology Based On Parallel Coordinates

The parallel coordinates technique is one of the important data visualization means to present multi-dimensional data. The technique makes multi-dimensional data expressed on the surface. The analysis ways with parallel coordinates technique to deal with datasets have obtained a great progress, including brushing, exchanging coordinates, and abstraction, etc. These analysis ways have been applied to many fields of data mining, and played a more and more important role in the development of data mining technology. The work of the thesis is supported with the National 863 Hi-Technology Project "Domain-Oriented Data Analysis and Data Mining Technology". In the project, the author's research mainly focuses on the visual data mining technology. This thesis presents the research results of visual data mining technology based on parallel coordinates. The main contributions in this thesis are in two aspects: [1] A method of visual clustering analysis is improved by data visual technology of parallel coordinates. Utilizing the merit of common parallel coordinates and hierarchical parallel coordinates, the effect of visual clustering analysis can be improved largely. [2] A method of visual data classifying is improved by integrating visual technology of parallel coordinates with data classification algorithms. [3] A visual data mining prototype system, called KingVis, based on parallel coordinates was designed and implemented. The system consists of two modules: visual clustering analysis and visual data classifying, with respect to clustering algorithms and classifying algorithms, in which the visual technology of parallel coordinates is combined to obtain better visual data mining effects.