1. ¹Microsoft Research, Redmond, WA98052, USA
2. ²School of Electrical and Computer Engineering, Purdue University, West Lafayette, IN 47907, USA.
3. ³SSS Research, VisTracks, Lisle, IL 60532, USA.
4. ⁴Department of Computer and Information Science, University of Konstanz, Konstanz, D-78457, Germany
5. ⁵Department of Computer Science, University of California, Davis, CA 95616, USA

Correspondence: George Robertson, PO Box 177, Northeast Harbor, ME 04662-0177, USA. E-mail: ggr@microsoft.com

Received 27 May 2009; Revised 7 July 2009; Accepted 7 July 2009.

Abstract

The fundamental problem that we face is that a variety of large-scale problems in security, public safety, energy, ecology, health care and basic science all require that we process and understand increasingly vast amounts and variety of data. There is a growing impedance mismatch between data size/complexity and the human ability to understand and interact with data. Visual analytic tools are intended to help reduce that impedance mismatch by using analytic tools to reduce the amount of data that must be viewed, and visualization tools to help understand the patterns and relationships in the reduced data. But visual analytic tools must address a variety of scalability issues if they are to succeed. In this paper, we characterize the scalability and complexity issues in visual analytics. We discuss some highlights on progress that has been made in the past 5 years, as well as key areas where more progress is needed.