Next: Hierarchical Structure
Up: Structure-Driven Glyph Placement
Previous: Structure-Driven Glyph Placement
Ordered structure may be linear (1-D) or grid-based (N-D).
Good for detection of changes in the dimensions used in
the sorting
May provide clues into potential dependencies in the other
variables
Common linear ordering include
raster scan [46],
circular [26,
35], and recursive
space-filling patterns [27].
Figure:
Various placement patterns for linearly structured data (from
[26,27]), including raster, radial, and recursive raster.
|
Figure:
Profile glyphs of monthly economic time series data using linear
ordering and spiral layout (from SpiralGlyphics[35]). Dimensions
are (from left to right) Dow Jones average, Standard and Poors 500 index,
retail sales, and unemployment. Data for December radiate straight up (the
12 o'clock orientation).
Expected patterns include high sales and low unemployment in January
and the inverse in June. Uncharacteristicly high unemployment is visible
through much of the first year shown, which improves as other market
indicators rise in value.
|
Figure 7:
Star glyphs of remote sensing data (5 channels) using 2-D raster
ordering. Regions of similarity manifest themselves as regions of texture.
Second image shows a zoomed in region with high Uranium values highlighted.
|
Next: Hierarchical Structure
Up: Structure-Driven Glyph Placement
Previous: Structure-Driven Glyph Placement
Matthew Ward
1999-02-08