Visualizing Food 40 Ways
Some people keep a food journal because, according to nutrition experts, it's one of the simplest and most effective ways to lose weight. But for New York-based designer Lauren Manning, the "food" part of the food log was borderline irrelevant. What Lauren wanted to get at, in tracking 730 days' worth of chicken, french fries, and falafel, was a single, robust data set that she could use to explore multiple visualizing possibilities. Something like the dataviz version of the epicure's egg.
The result of Lauren's experiment, the Visualization Matrix, was recently installed as part of her Pratt Master's thesis presentation, generating rave reviews on the design blogs. Intrigued, we reach out to Lauren to learn more. (See all the photos on Flickr.)
V: What inspired this project, and how did you decide on the different views for your food data?
LM: Starting in 2009, I kept track of every day of my life in little Moleskine notebooks documenting about sixty data points for two years:
I was inspired by Nicholas Felton's Annual Reports but also interested in exploring the shortcomings of human memory and trying to push beyond what we can remember on our own. When researching topics for my thesis, I became most interested in the levels of effectiveness of data visualizations. There were a lot being made, but many were done so poorly that they were hard to understand, or worse, misleading. I began to formulate a structure to compare different types of visualizations to evaluate levels of effectiveness for different methods/techniques. After gathering a few case studies from major news events that were heavily visualized, I decided to create my own set of visualizations to push this further. Working with only one data set allowed me to push and pull the same information various ways for different visualizations. I chose to use my food documentation because it was such a rich data set. Instead of just quantity, the food had many layers of information that I could work with additionally like location, imagery, time, experiences, relate-ability to the viewer and other elements.
The different methods I used to visualize the food centered on several objectives for the project. I wanted to show a wide range of styles, methods, and techniques and also use the varying layers of the data set - quantity, time, location, additional context, etc. For the visualizations seen on the left side of the visualization matrix, only one set of data is used, like times food was eaten or colors of the food. As the visualizations progress the right side, they start to contain more and more layers of information with some having four or more. I also structured the visualizations from abstract to literal with pasta, for example, being shown from a small pink dot to a photo of a noodle.
V: What do you think are the benefits of being able to see multiple views?
LM: The idea to show multiple views of the same visualized data in an installation stemmed from Edward Tufteâ€™s Envisioning Information. In the book he discussed how two maps couldnâ€™t be truly compared because they were on different pages. Our brains have to rely on memory to compare from page to page, but when all of the visualizations are seen simultaneously a bigger picture can emerge and visualizations can be compared as groups of 2, 4, 10 or all 41.
One of the biggest benefits for my project specifically is simply being able to see many of the options that are available to the designer. Secondly, the viewer can quickly establish how a simple bar chart communicates differently than points of data color-coded by month, scaled by quantity on a map.
Also, various options within styles can be seen, whether it is versions of line charts, maps or photographic methods:
V: What other information would you like to link into your visualizations?
LM: As I was creating the many versions of data visualizations with this data set, I felt like I added in good bits of context here and there. I added the map as an element of context, only to realize later that it was essentially just another part of the data. It was then that I realized that there are really two types of context. I determined them to be â€śrelational contextâ€ť and â€śradical context.â€ť With relational context, the new information that is added -- things like additional elements of time or location -- functions as context and sheds new, or more, light on the data. However, relational context is still very directly related to the data already.
Adding radical context, on the other hand, can provide entirely new data and form new mental and visual relationships for the viewer. Some examples of radical context in terms of my food visualizations might include things like breaking down the food structure to a molecular level and then visualizing it, or adding a new set of data that is comparable but totally new to mine, like the food data of someone else in France, Africa or India. This context would allow for an entirely new dialogue to emerge. Since I plan to continue pushing this project further, I hope to partner with others who can provide these elements of radical context.
V: You included something called "experience cards" in your project. What were these cards about, and what information did you ultimately glean from them?
LM: The experience cards documented peopleâ€™s interaction with the installation and their feedback on the effectiveness of the different visualizations. The simple little cards asked the viewer to track their interaction with the project marking which visualizations they looked at first with a circle, looked at longest with a box, thought was most effective with an asterisk and thought was least effective with an â€śx.â€ť The results were all over the place. Some viewers marked only one visualization for each topic, while others marked each visualization with one or more of the topics.
Overall, the final conclusion showed mostly that effectiveness varies tremendously with the viewer, but that general trends did emerge. People often noticed the three visualizations below first.
One result that surprised me was how many people chose the pie charts as the most effective. Maybe it was familiarity, or maybe they really did find it most effective, but the two pie charts in the installation got pretty good results. I havenâ€™t quite finished tallying the results but Iâ€™m excited to see the exact numbers that emerge and hopefully do some visualizations of that data as well.
V: Did you learn anything surprising about your own food habits from the 2-year project?
LM: Since I was documenting so many data points and always planned on this project being public, I kept a pretty scientific approach to the data. However, at the end of each month I totaled certain items for food, drink, locations, hours slept, emails, flickr and a few others. It was interesting to see how each month compared to the others and I always found this creation of new data from the documenting to be so interesting. My food consumption specifically wasnâ€™t really changed by the project because that was not something I was trying to do with it. For me, this was more about being able to see the bigger picture. One thing I did really enjoy was seeing my food habits evolve as I moved from the South to New York City and also as I became more aware of how to eat moderately healthy food on a grad student budget.
V: With Americans increasingly being urged to skip the processed foods in favor of fresh, sustainably grown varieties, what use could your visualizations have for mapping other people's diets?
LM: There is a tremendous potential for data and documenting to play a large role in shaping our future. With the ability to transform data into something more tangible and visible almost instantly with new technology, I think we will definitely see this playing a part in our awareness of our consumption of all sorts of things. Platforms like Daytum (developed by Nicholas Felton) already let users keep track of their data and interact with it in a visual way but I think this is just the beginning. Imagine a grocery cart that helps shoppers keep track of the healthy versus processed foods theyâ€™ve chosen by visualizing it on the spot or a rewards/coupon system that works based off of purchasing healthy, fresh foods over processed junk foods. Data driven systems are going to be more widely implemented in all applications and the change and awareness that can result from these new systems could be really wonderful.