The Research And Application Of Multidimensional Data Visualization In Data Mining

With the development of database technology and the popularization of database application, the quantity of data that is stored in computer is being huger and huger increasingly. People want to analyze the data so they can obtain knowledge or information, instead of just managing them. Information visualization technology is one of important implements to display data, which can discover the relation between information and latent characteristic. Multi-dimension data visualization is a focus content of information visualization field.First, this paper make a summary of information visualization technology and some conventional methods for visualizing multi-dimension data, having an introduction about the principle and characteristic of various technique. In process of visualization the arrangement of dimensions are lacks of guiding; so many knowledge and information will be overlooked. In this paper a arithmetic of dimension similarity is applied. In the first step, the similarities of all the dimensions are calculated, and then a similarity matrix is built with these values of dimension similarity. At last an arrangement of dimension is gained with the matrix and optimization arithmetic. When the quantity of dimension is too large, it is hard for user to watch and understand the data. So the information gain arithmetic is used. Entropy is applied to account the information gain, and then series dimensions that have a low information gain are deleted from the visualization. Thus the users can find the knowledge and rules more easily. Both these researches get good experiment results, so they are applicable.In the last part of the article there is a practice on basketball player data investigation statistics analysis system. The contents of analysis are multi-dimensional and user want to change the contrast items dynamically, demanding display the data from different aspects and getting accurate result rapidly. This system combined conventional data visualization method with parallel coordinate technology to deal with data availably. User can get the visual result by operating interactively and analyze the results of many kinds of situations under parallel coordinates, reducing their workload greatly.