In our final week of the program, our project managers are requesting a visual interface that focuses explicitly on usability, as opposed to machine learning or any other experimental functionality. Whereas event classification and prediction (i.e. determining whether an event is happening or is going to happen) was a predominant goal of the summer, we are now focusing specifically on representing data usably and authentically, by overhauling the interface and recalculating mobility metrics.
Building on the interface demonstrated in our final presentation, the final week 10 interface caters to faster, more refined queries, and has an arguably gentler learning curve. Using a painting metaphor, the interface encourages the user to paint calendar days, and then observe how the corresponding colors compare over time. By request, a threshold slider is also introduced, so that even the least used routes between buildings can be viewed if desired:
Whereas the grey lines and circles represent the average usage and mobility measures across all days, the colors represent the specific days that the user selects. The legend in the bottom right helps for reference.
By the end of the week, our primary objective is to feed data sets unrelated to WiFi into our API, like weather and campus-wide event times, so that the interface user can see how events correlate to changes in WiFi behavior. This will allow third parties to make formal declarations about how certain events influence campus mobility, and present tailored policies thereafter. Here is an example using rainy days:
