From CS294-10 Visualization Fa08

Jump to: navigation, search


Good Visualization


Online tracking of hurricane Hanna, www.wunderground.com/tropical, taken at 11am EDT Wed Sep 3 2008


This visualization is extremely well designed because it provides massive amounts of information in an easily understandable format. The color choice for the range of storms can be understood even without reading the map legend since it follows from a relaxed light blue, to a dark red and further to a a vidid light pink(which offers perfect contrast to the blue background of the ocean/gulf). The information presented in the leftmost corner also provides exact data about the storm in a succinct way.


This visualization presents allot of information bunched up in a small space:

  • Quantitative information about the hurricane is presented in the top left corner describing location, maximum wind, gusts and minimum pressure
  • Ordinal information is presented in the bottom left corner describing the range of hurricanes and using color to encode their ordering (controversial but it uses 'learned' colors rather than random)
  • Quantitative information for each category hurricane, tropical deppresion/storm
  • Quantitative information represented in the positioning of the assumed hurricane path over the next 3 days

Bad Visualization


Map of Missouri, Show me a showdown, Aug 28th 2008 | JOPLIN AND ST LOUIS, From The Economist print edition


This map at first glance looks to be a good visualization because it shows how each of the different local parishes voted in the 2004 election but that impression quickly disappears when you look at the procentage with which Bush won the vote over Kerry. From the map the percentage looks like it should be somewhere in the range of 85-90% for George W. Bush but when you look at the actual numbers you see that he actually won by 53.3% to 46.1% over Kerry. When you look again at the map you can see a hint of why that is so if you realize the portions of Missouri that voted for Kerry are the parishes that are close to or include Kansas City and St.Louis, big towns with big populations with respect to the state. The map is misleading because while most parishes did vote for Bush in the 2004 elections those parishes were much less populated than the ones which voted for Kerry.


The data presented in my bad visualization presents:

  • nominal values for each county (who won the county; red - Bush, blue - Kerry) (may be considered ordinal also depending on argument)
  • a small set of quantitative values in the percentages with which George W.Bush won the state and the margin of votes by which he won it.
  • Some geographical information is also presented in the shape of the counties of Missouri(quantitative values)

This visualization was built from a much larger data set of values with only a handful of information being extracted so in order to fully represent the data I found:

Redesigned Visualization



The visualization has the following improvements over the old version:

  • the population of the counties is now shown with darker tones representing a larger voting population and with a logarihtmic increase in the size of the categories
  • the final results of the election in Missouri are now shown in a large pie chart to directly show the winner and the margin by which he won (in percentages)
  • each county has it's own small divided pie chart representing the percentages of votes won by the 2 candidates (other votes are ignored in the smaller pie charts)

[add comment]
Personal tools