Interactive Visualization of Large Data

• Activity Number: 472
• Type: Invited
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Peter Zhi-Yuan Wang*+
• Companies: Continuum Analytics
• Keywords: visualization ; graphics ; python ; web

Abstract

Traditional 2D plotting systems often do not consider the challenge of visualizing large datasets, where the number of points far exceeds the resolution of the output device. At best, naive downsampling or crude binning is applied, which can remove much of the texture and obscure interesting outliers. In this talk, we will present Abstract Rendering, a new, structured approach for the visualization of large data which uses a simple but fundamental shift in the traditional graphics pipeline. We will demonstrate applying the AR technique to visualizing large scatter, line, geographic, and matrix plots, and discuss how its architecture is fundamentally suited to supporting flexible, interactive visualization of large datasets in a browser.