Visual Representation of Data

1
Visual Representation of Data

• We represent data visually to ease and/or enhance
  our understanding of the data
• A table of numbers can be difficult to understand

2
Visual Representation of Data

• We can make a graph of the data and we can
  perhaps see trends more easily

3
Visual Representation of Data

• Our graph shows the trends in the numbers more
  easily than studying the table of numbers
• The graph lacks meaning something is missing
• The numbers havent been associated with the
  world
• They have no Context

4
Visual Representation of Data

• Context components
• Title
• Axis labels
• Curve differentiation using line quality (dots,
  dashes, thickness) or color
• Legend
• Text annotations

5
Visual Representation of Data
6
Visual Representation of Data

• We can create graphs with 3 axes
• Visualizations of higher dimensional data sets
  require Context to give meaning
• Scale
• Orientation
• Labels
• Have freedom to choose viewing angle, set
  lighting, etc. to highlight features.

7
Visual Representation of Data
8
Visual Representation Case Study

• Model a storm that took place on April 3, 1964
• The storm spawned a tornado that killed 7,
  injured 111 and caused 15 million damage
• The storm crossed Oklahoma and Texas for 2½ hours
• Initial values for the simulation are provided by
  measurements of temperature, pressure, moisture
  and wind velocity taken at the origin of the
  storm

9
Case Study Severe Storm

• A Study of a Numerically Modeled Severe Storm,
  Robert E. Wilhelmson, Brian F. Jewett, Louis J.
  Wicker, Matthew Arrott, Colleen B. Bushell, Mark
  Bajuk, Jeffrey Thingvold, and Jeffery B. Yost,
  International Journal of Supercomputer
  Applications, 4, (1990) 20-36.
• Work done at the National Center for
  Supercomputing Applications (NCSA), University of
  Illinois at Urbana-Champaign
• Shown at the 1989 SIGGRAPH conference

10
Case Study Severe Storm

• After the Storm Considerations for Information
  Visualization, M. Pauline Baker and Colleen
  Bushell, National Center for Supercomputing
  Applications, University of Illinois, Technical
  Report TR029, (January 1995).
• Work done with Edward Tufte
• Visual Explanations, Edward Rolf Tufte, Graphics
  Press, Cheshire Connecticutt (1997).

11
Case Study Severe Storm
12
Case Study Severe Storm

• In the original simulation
• the thunderstorm placed in a heavily gridded
  black-and-white background
• the computational space was marked by a dark
  shadow box
• this was done deliberately to emphasize that this
  was a simulation (focus is not the cloud but the
  computational space)

13
Case Study Severe Storm

• Questions
• How big is that cloud?
• What direction is it moving?
• What are the dimensions of the grid?
• The context of the visualization can be improved

14
Case Study Severe Storm
15
Case Study Severe Storm

• In the redesigned visualization
• the grid and and the shadow box are
  de-emphasized, using shades of gray with
  just-noticeable differences.
• downplaying the grid and backdrop suggests that
  the important player in the visualization is the
  cloud, which is exactly right.
• more contextual information was also added.

16
Case Study Severe Storm

• Labels provide qualitative information
• Axes provide spatial information about the
  direction of the storm's travel
• Note that the vertical scale is multiplied by 2
  to enhance the inner workings of the cloud

17
Case Study Severe Storm
18
Case Study Severe Storm

• The passage of time is indicated by a timeline at
  the bottom, with thumbnail-sized snapshots and an
  advancing red line.
• This gives the viewer
• an indication of the overall duration of the
  storm,
• some clues about how the storm cloud will develop
  over time,
• and a clear indication of where we are in the
  sequence.

19
Case Study Severe Storm

• The new visualization makes it easier for the
  viewer to understand the information being
  presented
• We want to answer the quantitative questions
• How Many?
• How often?
• Where?
• How much?
• At what rate?

20
Visualization

• We can design visualizations which do not answer
  the quantitative questions
• These contain little useful data and perhaps lots
  of glitz
• A visualization like this can be fun