Research On Visualization Model And Its Applications In Data Mining

Data mining is the automated or convenient extraction of patterns representing knowledge implicitly stored in large databases. Visualization is a sort of technology that expresses data objects, and in data mining visualization is mainly used in data objects and data mining process and so on. Visualization techniques are needed to deal with the output of high-volume data sources in data mining.Visualization techniques are mainly used for the data visualization in data mining. For the analysis methods and processes itself, the visualization is often not effective. The relationship between visualization and data mining is often loose.To apply visualization in data mining, or to establish visible data mining method is a cross research subject about visualization and data mining. This type of research is required to be established on reasonable acknowledge basement. On one hand, it requires to analyse the theory and technology basement of this method; on the other hand, it also requires to consider the visualization characteristics of the property of data mining subjects and the observer's awareness of visualization characteristics.Two aspects are mainly needed to be considered during applying visualization to data mining. One is the separability of the mining algorithm process, that is, to split the process of the mining algorithm to inaffect the result of the data mining. The other is to determine the key factors in the mining algorithm and measuring standard, and find out their influences to the result of data mining.To explain the visualization result of multiple dimensions data in a better way, the process of visualization's objects should match the process of the data objects, so it's able to draw necessary measurement index into visualization application. To draw into proper measure index in quantification contributes to improve visualization techniques, and design applicable measuring index, and establish functional evaluation model.In analysing and processing the visible objects, visualization techniques based on measuring index can get help from proper mathematic method to model and evaluate, contributing the research and application of visualization of data mining.The application of the data mining algorithm based on visible measuring index provides a visible data mining method. In the process of data mining, visualization technology contributes to find out data characteristics. To use measuring index as a evaluating index can improve the result of mining by improving the algorithm parameter and process.This dissertation focuses on formalization expression of data and data mining visualizations model, Measurement model of visualization techniques and some application fields.Research of the dissertation is concentrated on the following aspects:(1) The characteristic and method of visualization techniques applications in data mining are analyzed. A general mathematical form of data and data mining process visualization is presented, namely convey model, association visualization model, association statistical visualization model, and process visualization model.(2) The paper presents a concept that describes characteristics of data attributes as the influence degree, and its icon-based visual expression and visual computing. The concept and method are used into visualization of data and statistical information of data attributes. The results show that the method is simple and valid for data visualization and data analysis.(3) The paper proposes a parallel coordinates-based measurement model and index system, proves some related properties and conclusion. We set up a theory of measurement model, and give an application case with K-means algorithm. The results show that the measurement model and index system are valid for data and data mining visualization.(4) The paper describes an approach that uses octree technique into data mining system, and designs a 3D data exploration environment. We discuss an approach to visual associations based on the parallel coordinate technique, and a way to visualize the multi-dimensional data with polygon.