Footprint Splatting and Density-based Parallel Coordinates for Interactive Visualization of Large Data Sets

M. Sanver (UAE) and L. Yang (USA)

Keywords

Footprint splatting, density-based parallel coordinates,large data visualization

Abstract

Many visualization techniques are not scalable to the size of input data and lose their effectiveness on large data sets. We present extensions to two visualization techniques, namely, volume rendering by footprint splatting and parallel coordinates, to visualize large relational data sets. Large relational data are aggregated into high-dimensional hyper rectangles, which serve as inputs to data visualization. The footprint splatting technique is extended such that foot prints of high-dimensional hyper-rectangles are generated by a two-step bootstrapping method. The parallel coordinate technique is extended to density-based parallel coordinates which visualize high dimensional hyper-rectangles as translucent bands across the parallel coordinates. We provide ļ¬ltering, brushing, zoom-in, and zoom-out func tions to interactively explore large data sets. The density based parallel coordinates technique greatly reduces clutters in traditional parallel coordinate visualization of large data sets. These extensions are implemented in a visualiza tion tool, n23Tool. Experimental results are presented to illustrate the usefulness of these visualization techniques in visual analytic of large data sets.

Important Links:



Go Back