A MONTH IN AN ANIMAL SHELTER
date Spring 2018
course Thesis – Exhibition | Northeastern University
professor Paul Kahn
THE EXHIBIT
Created for my masters thesis exhibit, this project gave me the opportunity to apply my thesis research on appeals to emotion in visualization and to visualize a population I care strongly about. This work explores one month of animals in the Animal Rescue League of Boston's three shelters. My exhibit is comprised of two parts: the interactive visualization in the foreground and the button pins in the background. The button pins served as takeaways that highlighted data points in the dataset visualized, putting faces to the data. I used Photoshop to design the pins and the D3.js library to create the web-based visualization.
THE VISUALIZATION
The visualization allows for exhibit-goers to look inside an animal shelter and explore its population in different ways. An important concept of this work was to represent the mass as individual animals. Therefore, the primary view of the visualization takes the form of a disaggregated chord diagram, illustrating the flow of animals in and out of the shelter, grouped by their intake type and outcome type. This view animates in, starting each animal's path based on their intake date and drawing the path for the duration the animal was at the shelter during the month. Since it's difficult to discern each animal's duration after the animation, I created a secondary view to illustrate the how long the animals were in the shelter. The animal paths transition between these two views smoothly, enabling users to follow certain populations during the view change.
Supplementary information in the right-hand column includes summary information, a bar chart of flow or histogram of duration (depending on the selected view), buttons for animal types, and buttons for selected animal stories, which correspond to the animals on the takeaway pins. The buttons allow for sub-populations to be highlighted by the user; the visualization view, summary information, and bar chart respond to the interactions. In the primary view, intake and outcome types can be moused over to highlight those sub-populations as well.
Since this work was developed as an interactive exhibit piece, I created a fun idle page that is triggered after 5 minutes of no activity, which includes animal prints animating across the page.
To interact with the visualization, click here.
THE PROCESS
My initial exhibit ideas had several other components. My first plan is represented in the 3D paper model below. This plan had an animated visualization on a large flat screen (a student can dream). The idea for animated dots was largely inspired by Fathom's athenahealth visualization. I also had an informative poster, a poster of button pins, my thesis book, and a tablet for an interactive version of the visualization. My second iteration simplified my idea somewhat by removing the poster and thesis book. This iteration is represented in the two storyboard images below. This iteration better clarified the elements of my visualization, but still separated the animated and interactive aspects of the visualization on two separate devices.
Ultimately, I went with a less-is-more approach and created a single visualization that first animated in, then allowed exploration through interaction. While all of my sketches illustrated the animals as dots, I found through data analysis and initial programming that representing the animals as paths was more suitable for this project. I came to this conclusion to allow for comparison from intake types to outcome types after animation and to encode duration into the length of the path in the second view. By focusing on just one month, the number of data points was suitable enough to move forward with individual paths using D3.
FUTURE IMPROVEMENTS
While the visualization was successfully implemented for the exhibition, there are a few improvements I'd like to make so the visualization is useful and successful long term. The first improvement will be enhancing the initial animation of the paths. There are some browser differences in the implementation and fluidity of the animation, and the fluidity decreases with more applications running on the device. The second improvement will be building interactions on top of one another. Current interactions highlight a sub-population without consideration of the previous highlight. By changing that, users can dig deeper into intersections of sub-populations, like all dogs who were owner-surrendered. Lastly, the duration view has room for improvement. The paths are currently positioned on the x-axis in an index order from longest duration to shortest. More thoughtful consideration to the sorting of paths, or the ability to re-sort paths, will help users get more out of this view.