Sunday, June 13, 2010

Post winning Tableau contest

Some of the things I experienced from winning Tableau visualization contest:
- It was pretty fun to see the search result on my name got over 10000 hits, the day after the press release
- The Fit & Fat viz appeared in Times Square New York
- Celebration from my colleague at DePaul (wine & cheese)
- Acknowledgement from the Dean and faculties of School of CDM
- Announcements in DePaul Facebook, DePaul Magazine, EM&M Newsletter
- Free trip from Tableau to Web 2.0 (San Francisco) with my husband 
- Radio interview with Charlie Hagger (KOMO Radio Seattle)
- Spoke at the kick-off meeting at the San Francisco User Group meeting (May '10)
- Spoke at the kick-off meeting at the Chicago User Group meeting (June '10)
- Interviewed by Dan Murray (blog coming soon)
- Speaking engagement with the Federal Home Loan Bank (mid July)
- Speaking engagement with the Atlanta Tableau User Group (Aug)


I'm also making new contacts, especially with those who are interested in data visualization: with folks in Tableau, Tableau experts, consultants, bloggers, from inside and outside the States. I think that is the most valuable award of all.



So why my viz won the contest? An internal source told me it was the extra analysis (the + and - factors, the two charts at the bottom) and the design (which comprise of a lot of things).
The dataset contain 90+ variables. It was a challenge to come up with a number of variables that are highly correlated with obesity rate. I'm glad the extra step paid off.  

As St. Vincent De Paul once said 'It's not enough to do good, it must be done well'.

Thursday, April 22, 2010

Supersized caskets

NOTE: This viz won the Data Visualization Contest, sponsored by Tableau and ReadWriteWeb. News release.

Obesity is a prevalent health issue and a growing epidemic in America. If you put it in a broader context, the average obesity rate in the US is the highest in the world (27.2), more than twice the worldwide average (14.1). The airline seats are too small for us, and so do the caskets.

The viz below gives us insights on who is living fit or fat in the US and why. The map shows that there is a regional connection between where we live and the rate of obesity, as the Midwest and South hold the highest obesity rates (represented by the orange colored bubbles). There is also a strong relationship between obesity and diabetes (bubbles that are orange are also larger bubbles).

The low-income family/community have higher prevalence on obesity than those with higher incomes. This is exemplified by the community characteristics chart on the bottom, which shows factors that are positively correlated with low or high obesity rates. The viz shows that those who have low obesity rate usually have higher income and dine in full-service restaurants, while those with higher obesity rates commonly have lower incomes, have limited access to grocery store or fresh food and eat at fast food restaurants. Black communities in the South also have a higher prevalence to obesity than Black communities in other regions.

Obesity is not an epidemic for certain communities only, but for ALL Americans. Why? Because the 'leaniest' state in the US (Colorado) still has a higher obesity rate than worldwide average.
Interact with the viz and see the waistband of your community and to what extent your community's profile compares to the national profile. There is also a potential for community leaders to use this viz to improve the health of their communities by dealing with the negative factors that have larger gaps compared to the national average.

How to use the viz:

  • To select region, use the checkbox option on the right.
  • To select one or more states, click on the bubble in the map or highlight multiple states by dragging your mouse. You could also use the drop down menu. The counties within the state will be displayed according to the state you selected. To release your selection, click on the anywhere on the map outside the US states or use the drop down menu.
  • To view profile by county (ies), click the bubble or highlight multiple bubbles using your mouse, or use the drop down menu. To release your selection, click on the axis (diabetes rate) or use the drop down menu. Remember to release your selection on County to return to overall view. (I'm yet to figure out how to do this more effectively)
  • The profiles (graphs on the bottom) will change according to the area you select (region, state or county). The characteristics are sorted by the difference between community profile vs overall US profile.








Horizontal version of this viz (fit in one screen & no scrolling) 

Note: The original entry to the contest was the horizontal version, where the entire viz fits in a screen. This viz was then re-structured by Tableau so that it would fit in a blog site which is vertical and narrow. I'm not very keen on the vertical version since it's against one of the rules in designing a good dashboard where users should be able to get a good glance of the dashboard in one screen, with no scrolling.