The Research On Visual Data Mining Technology Based On Parallel Coordinates Method

The data mining technology's complexity, the user is very difficult to master, the result which obtains is also very difficult to explain. But the people are acceptable and intelligible to the graph and the image.The visual data mining technology is on the rise and presents the broad application prospect.The parallel coordinates method is one of the important data visualization means for visualizing high-dimensional geometries using a two-dimensional display. It is intuitionistic to display the multi-dimensional data on the two-dimensional surface and intelligible to analyze data. The analysis ways with alternation and distortion technique to deal with datasets on parallel coordinates have obtained a great progress, for example brushing, abstraction, the dimension control, zoom control as well as the hierarchical display mode and so on.These technologies have provided the good premise and the foundation for visual data mining technology research based on parallel coordinates.The paper carefully researched visual data mining technology based on parallel coordinates method. In view of the parallel coordinates method and data mining technology characteristic and data mining visualization display mode, this paper designed and realized a parallel coordinates visualization prototype system. This topic research mainly includes the following several important aspects:First, the paper studied the parallel coordinates visualization,combined the alternation of parallel coordinates with the user, and more reasonably realized visualization alternately. The paper implemented many operate methods of parallel coordinates in software procedure, like brushing, the dimension select, rearrange axis, the dimension zoom, the data abstraction and so on.Second, the paper designed the data mining visualization hierarchical structure, took the visualization data mining module as a foundation, realized the concrete algorithm in the intermediate level, transferred interface way with the connection to realize the concrete algorithm and the visualization technology separation, and achieved software scalability and versatility. In the application layer the results of the visual mining was displayed and completed information transmission.Finally, the paper finished the parallel coordinates visual data mining prototype system.The system combined closely the data mining algorithm with the parallel coordinates visualization graph demonstration, programmed and implemented two clustering algorithms and one classification algorithm. Based on the visualization result and the corresponding data output, the problems which existed in the concrete data mining were analyzed and researched. The problems which appeared in the algorithm were carried on corresponding processing, and one algorithm was improved simply.Data mining realization, through the parallel coordinates prototype system, manifested parallel coordinates visualization the great significance which brought to the data mining technology, analyzed the parallel coordinates method applicability thoroughly as well as merit and shortcoming which compares with other visual methods.