General CS Event
Visual Data Analysis: Perceptual Foundations and Applications
May. 06, 2016 @ 11:00 am
Florida International University
IBM T.J. Watson
The explosion of digital data in size and form across domains surged the demand for data visualization. Visualizations facilitate data analysis by leveraging visual perception to support exploration and reasoning. In the first part of my talk, I'll present two interactive tools for visual analysis of complex data. I'll demonstrate how both tools help domain experts explore patterned structures in data effectively by reducing the complexity with novel visual encoding and interaction techniques. I'll recall Tukey & Wilk's analogy between data analysis and running experiments, and conclude the first part of my talk by proposing visual analysis like doing experiments as a high-level design guide for developing visual analysis tools. In the second part of my talk, I will consider what-in measurable terms-constitutes a "good" visualization. I'll propose visual embedding as a new model of data visualization, claiming that good visualizations should perceptually preserve structures in data. I'll then introduce perceptual kernels, distance matrices derived from aggregate perceptual judgments, to further operationalize the visual embedding model. I'll discuss how to best elicit perceptual kernels by presenting results from a large-scale crowdsourcing study we conducted. I'll finally demonstrate how perceptual kernels can be applied to improve visualization design through automatic palette optimization and by providing distances for visual embedding.
Çağatay Demiralp is a researcher at IBM T. J. Watson Research Center. Çağatay's research focuses on two themes: Extending perceptual and theoretical foundations of data visualization and developing new tools and techniques for interactive visual data analysis. Recently he was a postdoctoral scholar in Computer Science at Stanford University and a member of the Interactive Data Lab at the University of Washington. Çağatay obtained his PhD from Brown University.