Visualizing data with Protovis

In this month's Communications of the ACM, Heer, Bostock and Ogievetsky provide an eye-opening overview of the latest visualization techniques in A Tour Through the Visualization Zoo. The graphs and figures in the article were all produced with Protovis, an open-source library developed at Stanford for producing visualizations using JavaScript and SVG.

I've reproduced a few of my favorite visualizations from the article below.

The first is an index chart which shows the relative change in stock prices of several tech companies over time. The gain/loss factor is based on a fixed point in time (the red line). You can see some correlation between the stock prices, and clearly Apple is the overall winner. The online version allows you to change the location of the red line.




This is a choropleth map showing obesity rates in the US in 2008. The online version allows you to slide the date back to 1995. As you move the dates forward, you can see how the US population gets fatter, with the highest proportion of obesity rates focused in the South. There's a saying here in Arkansas: At least we're not Mississippi!




When you view the obesity rates with a Dorling cartogram, you see that Arkansas actually has fewer obese people than Colorado!




The next couple of graphs show how to visualize networks. Both visualize character co-occurrence in chapters from Victor Hugo's novel Les Misérables. The first is a force-directed layout. The node in the middle is, of course, Valjean. The online version allows you to pull nodes around and watch them re-arrange themselves.




The nodes can also be rearranged in an arc diagram as shown below.

Apple vs. Google vs. Microsoft War Chart

This infographic by Shane Snow beautifully illustrates the epic battle that Apple, Google, and Microsoft are engaged in. From ebooks and phones to office and email, all three giants are battling for more users.

Although many of these services are free to the end users, there are billions of dollars in revenue at stake. Click on the map to zoom-in.

Map of Science

This past summer I worked with the Digital Library Research and Prototyping Team at LANL. One of the projects they were working on, a "map of science", was just featured on the Wired Science Blog. The graph below is based on a massive collection of scholarly usage data (click on the image to get a more detailed look at it). Basically, the graph shows how users accessing scholarly work in one field (e.g., Cognitive Science) may also access work in another related field (e.g., Sports Medicine).



You can find more information about this work in:

Clickstream Data Yields High-Resolution Maps of Science by Johan Bollen, Herbert Van de Sompel, Aric Hagberg, Luis Bettencourt, Ryan Chute, Marko A. Rodriguez, Lyudmila Balakireva. Public Library of Science ONE, March 11, 2009.

Bible visualizations

Check out the many Bible visualizations created by Chris Harrison. The one below shows 63,779 cross references from one verse to another, represented by colored arcs. The chapters from the Bible make up the horizontal axis, and the vertical bars off the axis are the number of verses in each chapter (that long one in the center is Psalms).

God's Word never looked so beautiful.


Thanks to Jeremy Bullock who pointed these out to me.

Facebook visualization: Sharing around the World

Allan pointed out this beautiful visualization that from Facebook showing the interactions of users around the world. Watch comets orbit the Earth as friends post messages, write on walls, and comment on photos. Very nice.

Fav5

My pick of the week's top 5 items of interest:

1. Neil McAllister asks, Should computer programming be mandatory for U.S. students? A related article discusses how math skills are suffering in the US, mainly because US culture does not value mathematics and discourages females from pursuing it.
   
   


2. Do you know what Web Science is? Read Nigel Shadbolt and Tim Berners-Lee's article Web Science: Studying the Internet to Protect Our Future.
   

   This new discipline will model the Web’s structure, articulate the architectural principles that have fueled its phenomenal growth, and discover how online human interactions are driven by and can change social conventions. It will elucidate the principles that can ensure that the network continues to grow productively and settle complex issues such as privacy protection and intellectual-property rights.




3. New visualization techniques are being developed at the University of Utah to understand polling data. What struck me was the similarity between their donuts with the ones I developed to summarize reconstructed websites.



4. Leave it to the Japanese to develop a robotic girl that looks like something straight out of a horror movie.
   
   
   Uncanny valley?
   
   


5. This is hilarious. I've never seen more moronic comments than those on YouTube.
   
   
   

Wordle word clouds

Very cool visualization called Wordle that I found via Allan White's blog. Here's a word cloud I created using text from my blog:

I guess you could argue I'm a little fixated on Google.  

Fav5

My pick of the week's top five items of interest:

1. If you are attending Freed-Hardeman University, Oklahoma Christian University, or Abilene Christian University this fall, you may be getting a "free" iPhone or iPod (of course the students are still paying for it in increased tuition or technology fees). Several of our sister institutions are handing them out to incoming students so students can listen or record lectures, so campus-wide communications can be easily sent, and so teachers can take in-class polls. I'm interested to see if making these devices ubiquitous improves learning or just encourages students to phase-out during class.
   
   
2. GUI Blooper: Read about some usability issues in Windows Search.
   
   
3. Some MIT students were recently ordered by a judge not to present their report on hacking Bonston's subway system at this year's Defcon conference. The gag order has been dropped, but the students missed their opportunity to speak at Defcon. Instead of first sharing the security holes with the transit system and giving them time to fix the holes, the students were going to share their hacks at the conference but hide some of the details required to actually get free rides. This legal mess illustrates the many problems of turning security holes into public research.
   
   
4. Scour, a new meta-search engine with some social enhancements, is offering to pay users to use their search engine. You earn points for searching and getting others to search, and eventually you earn a Visa gift card.
   
   
5. A new visualization tool that helps experts understand flooding has recently been developed at Monash University. With their interactive tool, you can raise and lower the flood marks and watch the city drown.
   
   
   

code_swarm: Software project visualization

You need to see this. Ph.D. student Michael Ogawa at UC Davis has created an "organic information visualization" called code_swarm which animates contributions to a software project through time. Documents and source code fly in, hover around the developer's name, and gradually fade to black as commits become more rare. The soundtrack adds to the effect. Examples include Eclipse, Python, PostgreSQL, and Apache. (Thanks, Stephan.)

NameVoyager: Baby name visualization over time

Today I ran across this novel visualization tool called NameVoyager. It allows you to see how the popularity of baby names change over time. The data is collected from the Social Security Administration.

From the screenshot below, you can see all boy's names that start with E. Although Edward was very popular in the late 1800s, Eric started to dominate in the 1950s, and Elijah, Evan, and Ethan (our son's name, ranked #3 in 2007) have taken over since the 1990s.


My name hasn't fared so well (#6 in the 1880s, #262 in 2007). My wife's name went from a high of #13 in the 1970s, but it's since dropped to #105.

The creator of NameVoyager, Martin Wattenberg, writes about it in a white paper entitled Baby Names, Visualization, and Social Data Analysis.

BTW, if you are really interested in how your name affects your socioeconomic status, check out chapter 6 of Freakonomics by Steven Levitt.

Update

Another interesting name visualization tool is NameTrends.net. In addition to showing popularity of names over time, they also show geographical popularity of baby names. The map below, set at 1992, shows how Ethan was really popular in the northeast and north-central states before it became popular in all the states.

Wal-Mart growth: 1962-2007

This is one cool visualization. Watch as Wal-Mart grows like a cancer, beginning from Bentonville, Arkansas. (It's amazing to think the world's most powerful discount store started in humble AR...)

TouchGraph Facebook Browser

If you have a Facebook profile and want to see which of your friends are in your "inner circle", you might want to checkout the TouchGraph Facebook Browser. Below is screen shot showing the social connections between me and my "closest" 50 friends.


The connections between nodes is computed using betweenness centrality, a measure of a person's importance in a social network. This measure gives more weight to friends that branch other cliques. I'm not sure what the size of the nodes relate to.

The colors of the nodes indicate cliques/clusters - cliques have many friends within a group of friends but few connections to others outside the group.

• The purple group are mostly friends who were in the social club Knights with me in college.
• The red group are mainly active Harding students, employees, or spouses. Becky is in this group.
  
• The green group are people I know from when I lived in Denver years ago. My sister Sara is in this group.
• The blue group are other friends from college.
  

The large green circles are the networks my friends belong to. Most of the people in this graph are in the Harding or Little Rock network.

When I boil it down to only 4 friends, I'm left with Becky (78 friends in common), Hank Bingham (48), Jim Miller (45), and Mark Elrod (45). What would be interesting is if these connections could be recomputed based on communication levels... which friends do I communicate with the most?

Baseball and search engine queries

This is a really interesting visualization from Spatial Variation in Search Engine Queries, a WWW08 paper by Backstrom & Kleinberg (Cornell) and Kumar & Novak (Yahoo!).


What you are seeing are the "spheres of influence" for various Major League baseball teams. To create this graph, the authors mined the search engine logs from Yahoo! and plotted where the most popular baseball team queries originated from (based on IP address).

The result is a geographical break-down of each team's fan base. The authors point out that the boundaries even follow state boundaries closely: "For instance, in Michigan, across the lake from Chicago but far from Detroit, it is the Tigers, not the Cubs who have the largest following."

It's also interesting to note there are some places like Arkansas where there is no single winner. That's what happens when you live in a state with zero professional teams. Oh, how I miss my childhood years living in Denver...

Cool visualization of the day

I came across this today. It's a figure from The Political Blogosphere and the 2004 U.S. Election: Divided They Blog (2005) by Lada Adamic and Natalie Glance showing a citation network of the liberal and conservative blog communities.


There are 676 liberal (blue) and 659 conservative (red) blogs represented, and the size of each blog point reflects the number of other blogs that link to it. Blue links connect liberal blogs, red connect conservative blogs, yellow links go from liberal to conservative, and purple from conservative to liberal.