Research On Multi-dimension Data Visualization

In scientific research and daily life, we always face with multi-dimension data which can indicate lots of detailed information. Researchers paid unremitting efforts on how to denote these data directly and explore interesting information from the data. By transformation data information into intuitionistic, graphic, variational physical phenomena and variable, visualization technology can present us things and phenomena which are sightless in traditional sense. Therefore, this paper used visualization method to study and analyze multi-dimension data, which has been proved to be an effective way to deal with the data.Because of various methods used in visualization, we utilized corresponding method of data visualization according to the different data sources which were involved in our study. Firstly, according to the feature of theory of failure rate which obey the rule of bathtub curve in shape, the paper established the failure rate regression model based on life data. By means of using segment Weibull distribution to fit bathtub curve we can achieve the life data visualization based on density estimate. Meanwhile, we can calculate the parameter estimate of segment Weibull distribution. In order to display multi-dimension data in low dimension space, we used project pursuit method to reduce the dimensions of the data, and projected them into the visible 2D or 3D space, then analyzed the relation of data by clustering analysis. Finally, we introduced radar graph and face graph to figure multi-dimension data as well as provide the method of drawing the picture. Also we integrated the Canonical Correlation Analysis and Principal Component Analysis into mining the relation of multi-dimension data.Using the given study methods, we may draw graphs in computer by programme. In order to program efficiently we can use TeeChart component. The component can provide us powerful charts which can create and display all 2D and 3D charts and graphs in an impressive OpenGL environment; therefore, it can satisfy our desire. The running results verified the correctness of our programs and validity of our research methods. With the rapid development of internet the paper provided data visualization methods based on browser/server mode, which could extremely broaden the application fields of our research.In the end of the paper, we discussed the application of multi-dimension data visualization as well as research directions recently which are statistical analysis in economic data and decision analysis in e-government system. We implemented multi-dimension data visualization successfully by our research, which provided us direct data information. By the given analysis methods we can discover the relation of data as well as form a solid foundation for achieving data information deeply.