When I was working as a part-time UX designer at Institute for People and Technology, a group of research labs at Georgia Tech that do computing research in partnership with industry, I designed a data visualization interface to allow users to explore building occupancy data for the Georgia Tech's campus. 

This was an experimental project and I was the only person on the team, so I handled both the research and design part of the UX process. At the end of the project, I delivered a high-fidelity design concept detailing the data visualizations, time filters, and navigation design. I also helped prioritize features for development using user research insights. 

The user research effort helped uncover possible use cases that led to new partnerships for the lab. The new design when implemented will reduce the workload on the team since this data would be accessible online and the team will not have to pull it from the back-end every time someone requested it.

Before jumping into the design, I made sure I understood the data set thoroughly. So I talked to the advising professor to understand the different data fields, the technology with which the data was collected, and its granularity and limitations. I also explored possible data representations through quick sketching. This helped me further understand what the data was and what it was not.

I understood that the Wi-Fi infrastructure around Georgia Tech campus has several access points which detect the user's device and connect it to the 'GT-wifi' (the name of the Wi-Fi network). To connect to GT-wifi, users have to log in with their GT id and password. The Wi-Fi infrastructure, hence, knows who is logged in where and when. By aggregating that information, the occupancy of a building can be estimated without violating anyone's privacy.

The professor was interested in exploring the use of this data for increasing operational efficiency on campus. So, I interviewed 6 different operational units on campus to understand the use cases they envisioned for this data.

Based on these findings, the advising professor and I prioritized the use cases that we should focus on. We rejected ones that were too complicated for us to tackle, would not benefit from the data we have, or would require a unique dataset that most other use cases did not. The professor also shared with me graphs from another application that was using the same data set to see if we could serve these use cases with those visualizations to start with. 

Using those graphs, I created wireframes representing data at three levels of granularity - campus, building, and floor. The line graphs are screenshots from the existing application. The campus map and floor map visualization are new concepts. 

I showed these ideas to two users and realized that the campus level representations would not be useful. I also needed to add the option to look at the data at different time granularities. So, in the second version, I included data at building and floor level, across day, week, month, semester, and year. I also included average data trends that could be filtered by days of week and hour of day as that would serve several use cases.

I showed the updated wireframes and created an InVision clickthrough. I showed these to three more users. I explained the information represented and asked them if they thought it would be useful and, if yes, in what way and what action they would take. From those insights, I understood which of these visualizations would be useful for which use cases. This helped us prioritize features for the development stage. I also got feedback on smaller details, for example, labeling time as am/pm instead of military time and expanding 'min' to 'minimum' as that might be misread as 'minutes'.

Due to funding constraints, this project did not immediately go into the development stage. But after the project finished, for my own practice, I created visual mockups and shared them with the professor for future use.