ASA Sections on:
Andreas Buja, Bell Communications Research
John McDonald, John Michalak, Werner Stuetzle and Steve Willis
University of Washington (1991)
In this video the authors discuss two principles of interactive data visualization: 1) Focusing and 2) Linking views of data. Focusing refers to the methods for controlling individual views. Linking refers to methods for relating multiple views to each other. Examples of focusing include controls for zooming, panning, projecting. Examples of linking are linked color brushing and linked identification.
The stills above are from the video:
Top left: Two linked views showing time series data. The lower view shows a raw time series plot with a highlight brush covering a small time interval of observations; the upper view shows a linked lag plot with the corresponding observations highlit.
Top right: Two linked views showing data from a speech experiment where observations of the 11 vowels of the Dutch language have been replicated 6 times. The lower view shows a representation of the experimental design; the upper view shows a 2D projection of frequency representations of the pressure waves.
Bottom left: The same Dutch vowel data in an Andrews plot. The various vowels form strands of parallel curves. Each vertical section through the curves represents a 1D projection of the multivariate data. The curves act as links between a 1-parameter family of 1D projections. [To prevent a misunderstanding: These curves are NOT the pressure waves, but a sinusoidal representation of the multivariate data!]
Bottom right: Two linked views of multi-spectral brain scan data. The left view shows the actual cross section. The right view shows a linked 2D projection of 3D spectral space.
The systems shown in this video were developed on Symbolics Lisp Machines during the 1980s. The first three images are from Buja's "Data Viewer" system, a prototype and in some ways a predecessor of XGobi . The last image is from McDonald's "Antelope" system.