Data scattering algorithm for very large multivariable data visualization

Data visualization techniques are very important presently. Techniques such as data scattering, 3D vector glyphs, vector warping, and hedgehogs are used for multivariable data representation. Data represented by these techniques are more reasonable than current techniques such as histogram graph or table, but they take a very long time to plot data. In this paper an algorithm which quickly creates very large data scattering is proposed. Data scattering represents each element of data by a sphere in 3D space. For example, 3 axes of 3D space, and 3 components of a color and the radius of a sphere are used to present data of seven variables. Since meshes or polygons are not required, this algorithm saves time and memory. IBM PC compatible with Pentium Pro Processor 200 MHz, 32 Mbyte RAM and Windows95 is used in our experiments. The numbers of data are from 1,436 to 232357 and computation time is from 0.05 to 5.44 second.