From CS294-10 Visualization Fa08

Jump to: navigation, search

Good Visualization

Image: fishsm.png

Description: This visualization illustrates that the same fish has different attributes depending on where it lives and what predator resides in their specific environment. Information is more general at the top of the visualization and increases in granularity as the eye moves down It starts with the different environments where you find the fish along with their predators. It moves to a detailed view of the physical differences in adult fish, to differences in the larva, and finally the genetic differences. It provides a complete picture of the differences and why they matter in a very digestible format without being incomplete or having extraneous information.

Source: Scientific American, May 2008, Page 77

Deconstruction: My good visualization is more of an information visualization than a data visualization. Position on the x-axis indicates what environment the fish lives in, deep or shallow water. Position on the y-axis indicates the view level or magnification. At the top, the visualization shows the fish in its environment. As we move toward the bottom of the document, the fish is isolated, then the larva is examine and then the genetic differences. The visualization provides a very succinct way to quick deliver the differences that have developed in a fish based on its environment.

Bad Visualization

Image: BadVis.png

Description: This visualization is illustrating that tooth decay is decreasing in both countries where water has fluoride added to it as well as those that don't. While it does depict that tooth decay is trending downwards in both sets of countries, it over complicates communication of the message while concealing subtleties that the visualization could possibly reveal.

Color is used to indicate fluoridated vs. unfluoridated with greens representing unfluoridated countries and blues/violet representing fluordiated countries. There is no discussion of the individual country trends, so in this context, these individual colors may be unnecessary. Alternatively, in order to have made the examination of individual country results easier, thinner lines that more clearly allowed viewing of intersection points.

Source: Scientific American, January 2008, page 65

Deconstruction: The data model is statistical. There are four variables encoded in this visualization.

  • Time is encoded on the x-axis. Time is quantitative information and using position to encode quantitative information is appropriate.
  • The number of decayed, missing, or filled teeth in 12 year olds is encoded on the y-axis, again a quantitative variable appropriately encoded using position.
  • Six individual countries are plotted. Each country is represented by a different color, which is also appropriate as color is efficient at encoding nominal data.
  • Three of the countries plotted add fluoride to public water sources and three of them do not. Color is also used in an attempt to encode this information and is perhaps overloaded. Identifying the two groups of countries is as useful as the individual countries themselves do not express much in this particular context. We don't have any regional specific information, so whether or not the drinking water had fluoride added is more interesting.

Redesign: When I initially selected this visualization, I thought it was weaker than I do now. I think that some of why I thought it was weak at expressing information is because the data itself isn't very complete. For example, there is less data for some countries than for others like for Finland and the U.S.. Additionally, we do know that three of the six countries have fluoride added to the water. However, the description with the the visualization indicates that the other three countries may have fluoride added to foods, etc., but we don't have any specific information. Comparisons that can be made using the visualization between the fluoridated countries and the unfluoridated (water) countries would be more meaningful if we had that information . I think people often look to the visualization to reveal additional information that might not be explained in any writing with the graphic, but this visualization doesn't provide very much additional information, either because the designer didn't have it or they chose not to include it. It does support or reiterate the claim though that tooth decay is declining across the board.

I have redesigned the graphic with only the existing data that is present.

  • Option 1: The first option I considered is not using a visualization at all to illustrate that in countries both with and without fluoridated water, the number of decayed, missing and filled teeth is decreasing. Given the information that we have here, the statement itself may have been sufficient. However, the visualization may encourage additional exploration than the statement alone would.
  • Option 2: The second option I considered was largely retaining the original format of the visualization as the time and number of teeth encodings were appropriate. I made changes around the use of color.

V1: I made the two country groups each a single color. While you can no longer discern individual country information very well, I think this enables for better comparison of the two groups, fluoridated and unfluoridated. The fact that the two of the three countries using fluoridated water had lower numbers to start with is more obvious now. It's also more obvious that one of the countries using fluoridated water now has a higher number of decayed teeth in 12-year-olds than the unfluoridated water countries.
Image: BadVisR1_0.png
V2: In this version, I made the two country groups share the same hue value but varied the brightness for each country in order to preserve individual country information. I also made the plot lines thinner and removed the horizontal, dotted gray lines and replaced them with very thin white pixel lines that are only really noticeable if you need to extract specific data from the visualization. The thinner lines make it easier to see subtle differences between countries.
Image: BadVisR4_0.png

  • Option 3:

I plotted the data in a similar way as in the original plot and option 2, but filled in the area underneath the plot. For some countries the data spans nearly 30 years and others only 10. This can be considered a problem with the data, but I was interested in trying to represent this as it could provide some information (although, we don't know in this case) about how long the different countries have been keeping records related to the study or simply call out the fact that the data isn't complete and to keep that in mind when examining the visualization. I filled in the area underneath the plot in an attempt to convey the amount of data, based on time, present for each country while still preserving the visibility of the line slopes. Hue encoded the two separate groups, fluoridated and unfluoridated, and shade encoded individual countries. Countries are also labeled. Time and number of decayed teeth remained encoded based on position on the x and y axes respectively. This method obscurs information, particularly in the more recent data where all countries are converging. I considered trying to reveal the hidden information by reducing the transparency of the shaded areas. However, this created areas without meaning in the visualization, which further convoluted this relatively simple data.


  • Option 4:

I used a bar chart for option 3. The y position still represents the number of decayed teeth. Countries are arranged on the x-axis starting with the country with the highest number of decayed teeth for each of the two groups, fluoridated and unfluoridated. Each country has two bars and the relative position of each bar indicates whether the data is from further in the past or closer to the present. The bars are also labeled to indicate the specific year of the records. Shade also indicates the more recent data. I used only the first and last data points for each country from the original data. The total delta is the most expressive part of the data in the context of demonstrating that tooth decay is declining in both countries with and without fluoridated water. I also decided to do this because the data sets don't start or end at the same time periods and this provides a way to look at the data in a way that it looks more uniform.


Conclusion: I dislike the data set for this visualization more than the visualization itself. The visualization has several good qualities (position for quantitative data, color for nominal), but is a bit noisy. If the goal of the visualization is to demonstrate and support that tooth decay is declining in both countries with and without fluoridated water and nothing else, perhaps with the goal of encouraging further exploration, it does that.
After working through a few different designs, I find that Option 2, version 2 and Option 4 the best. I think both communicate the changes in tooth decay for all countries and clearly represent that decay is declining in both countries with and without fluoridated water. Option 2 makes the group to which the country belongs more clear. Option 4 highlights the differences between the starting points of the countries while making it clear that they all have similar end points in the present. It's possible that the new data is over encoded (label, shade, relative position) and that keeping the shading constant is a better design.

[add comment]
Personal tools