From CS294-10 Visualization Fa08
Source: The New York Times, Tuesday, September 2nd 2008
This image shows the path and general fallout of Hurricane Gustav. I was impressed by this visualization as it packs a considerable amount of data into a small space. Showing the path on a detailed map with a colored region of the hurricane's range gives a good context for the color-coded points showing water levels in a number of different places. One can easily see how cities outside the hurricane's wind-range are still effected by hurricane-caused high water levels—data that would not be immediately clear with just a chart of numbers. Additionally the inset of levees in New Orleans provides topical information and efficiently uses the space on the map.
This visualization shows the following data: A 2-dimensional map of the region near hurricane Gustav including state borders and names, primary cities, and a faint impression of the geography of the region, shown in light gray so as not to overpower the other information, the path of Hurricane Gustav over time as red connected dots to show specific lat/long data points and times (sometimes), the "estimated extent of hurricane-force winds" shown as a 2D yellow transparent overlay on the map, additional specific wind-speed numbers shown as text near the place where the wind-speed occurred, the highest water level reached in a number of locations shown both as a three-level ordinal variable denoted by differences in the hue/saturation of orange dots and as specific values shown next to the colored dot, and lastly miscellaneous data points with text and dotted lines pointing to specific lat/long locations.
Overall this visualization is successful in portraying the effects of hurricane Gustav's pass through the southern US. Converting the water height to an ordinal variable before color-coding the cities avoids the problems of using hue/saturation to denote quantitative values, and the fact that the highest value is a dark color points out unusual data points such as the cities in southern Mississippi with very high water despite being far from hurricane-force winds.
One problem is that only two of the dots in the hurricane's path show a specific time for that dot. With only two data points, it's not clear to me whether that means there are two hours between each dot, or if those just happen to be the dots the designer decided to mark. Additionally, I have an issue with the dotted lines that point from textual information to physical points. Their convoluted twists and turns make it unnecessarily difficult to see where exactly they point. This is exacerbated by the proximity of the different lines, making it easy to follow the wrong line.
Source: Wired Magazine, September 2008
This visualization shows the average drying time of a number of different chemicals. What's surprising is that while it shows a clearly quantitative value comparison, it uses arbitrary colors and eratically sized dots to show the data. It gives no impression whatsoever of the relative differences for the different chemicals. Interestingly, the wired webpage for this same dataset simply shows a chart, however even the chart isn't sorted in any way.
This shows very simply two dimensional data with seven nominal data points with single quantitative time values (except for two points, which have a range for a value). The visualization, however, is incredibly misleading. It uses color coded dots to indicate the nominal value of which type of material is being shown, which in itself is fine. However, it also uses perspective in its visual representation, which causes the 2D dots to be of differing sizes, but there is no connection between the placement of items in the perspective and their quantitative value. There is also no relationship between the arrangement of the values on the 2D page -- they seem to just be scattered on some sort of flat surface. This design completely fails to show the differences between the quantitative values, such as the fact that the highest value is 212 times larger than the smallest.
This dataset warrants a very simple bar-style chart, as there is so little data to show. I decided that having mixed ranges and specific values in a chart showing purported averages is misleading, so I simply took the highest value of the two values given in range. I made this chart in Adobe Illustrator, I used their automated chart-making tools, but then manually customized the resulting graph. I kept the color scheme used in the original graphic to differentiate the different chemicals, so the colors would supplement the placement of the bars. I marked the specific times directly on the graph to give a specific number in addition to the visual differences. This helped with the fact that Liquid Paper and Magic Shell both dry so fast they're barely visible on the same graph that shows the 48-hour drying time of asphalt.
I originally attempted to create two separate graphs to deal with the huge difference in value, but decided that two graphs was confusing—directly printing the values for the two invisible values sufficiently shows the numbers.