The CI Hosts a Data All-Nighter

For 25 hours last weekend, the University of Chicago offices of the Computation Institute looked more like a lock-in party. Attendees inhaled copious amounts of pizza and coffee, students shook off sleep or took a catnap in the wee hours of the morning, and new friendships and connections formed. But the this lengthy stay resulted in more than just sleep deprivation, producing interactive data visualizations for health care and crime reports, 3D-printable keyboards, a rap-song generator, and much more.

The event was the fall hackathon organized by Hack@UChicago, a student group dedicated to learning and sharing new ways to work with programming, computers, and other technologies. This year, Knowledge Lab's Eamon Duede and Hack@UChicago students Pete Vilter and Hunter Owens, who both work with CI research centers, brought the friendly competition and a full slate of workshops on data and computational tools to Searle Chemistry Laboratory, attracting a crowd of around 80 willing participants from across campus and beyond.

UIC graduate student and Urban Center for Computation and Data researcher Alessandro Panella leads the Introduction to Civic Data Hacking workshop.

UIC graduate student and Urban Center for Computation and Data researcher Alessandro Panella leads the Introduction to Civic Data Hacking workshop.

Knowledge Lab researcher Nathan Bartley demonstrates the NetLogo agent-based modeling language.

Knowledge Lab researcher Nathan Bartley demonstrates the NetLogo agent-based modeling language.

Before the hacking got started on Friday night, attendees could take their choice of short workshops taught by Hack@UChicago members, CI staff, and visitors from the Chicago Harris School of Public Policy and NEURO Club. Sessions exposed the audience to some of the datasets and platforms used in CI research, such as Plenar.io, the Global Trade Analysis Project, and provided introductions to working with civic data, such as city and county data portals, and environmental data, including what goes in and comes out of atmospheric, agriculture, and economic models. Other sessions focused on useful tools, such as the visualization package D3, agent-based modeling with NetLogo, and the Leaflet mapping library for Javascript.

Then, around 11pm, the attendees were set free to apply that new knowledge to the project of their choice, with the only rule being a deadline of 6pm the following day. While only a few determined souls stuck it out for the entire night, many students came back Saturday morning to work through the day and complete work started the night before. In total, 8 teams made it to the demo session Saturday night, with the grand prize of a Printrbot 3D printer given to what CI Director Ian Foster determined as the best project.

Pete Vilter of the Center for Robust Decision Making on Climate and Policy (RDCEP) and Hack@UChicago draws a world map during the Intro to Environmental Data Hacking workshop (led with Alison Brizius of RDCEP).

Pete Vilter of the Center for Robust Decision Making on Climate and Policy (RDCEP) and Hack@UChicago draws a world map during the Intro to Environmental Data Hacking workshop (led with Alison Brizius of RDCEP).

Foster chose from a diverse range of products, such as a machine learning algorithm for solitaire developed by Booth School of Business student Vinh Luong, an app created by undergraduates Sam Baugh, Alec Snyder, and Kyle Jablon that generates random but grammatically correct rap lyrics based on any seed word (such as “love” or “victory”) and adds a synthetic beat, and a “hello world” for the cluster management software Mesos developed by Center for Data Intensive Science staff member James Porter.

Two projects were singled out for commendation, aside from the grand prize winner. Honorable mention was given to undergraduate Shuwen Qian -- who also performed the heroic duty of supervising the overnight shift -- and his modification of the popular push notification app Yo for use with the UChicago shuttle bus system. His script allows users to set a particular bus stop and the length of advance warning, then pushes a text and audio “Yo!” notification to their phones when the bus is close to arriving. Qian hopes to make the app available to UChicago students soon, and may expand it to include CTA buses at a later date.

The runner-up project was created by Jake Walker, an MBA student at Booth. Walker pulled together disparate, publicly available data sets about Medicare payments and hospital quality, and combined them into an interactive map patients can use to choose between hospitals. The visualization depicted each hospital as surrounded by a circle, with color dictated by a “price” score, and size driven by hospital quality score, offering important information at a glance. Over the hackathon, Walker was able to create a visualization for every hospital in Illinois, despite a technical hurdle: getting kicked off of Google for exceeding the rate limit of its API for determining latitude/longitude coordinates from an address.

A screenshot from Walker's hospital data visualization.

A screenshot from Walker's hospital data visualization.

Walker said that building this visualization lays the foundation for more expansive work he hopes to do with a start-up company, which will combine health care data to help patients make decisions about their care.

“This project is something we would implement in a much more real way,” Walker said. “I think this will be very useful for patients looking for hospitals or a doctor’s office and want a good one that’s cheap. We’d like like to help make that decision.”

Grand prize winners Ben Huynh and Peter Xu also put together a useful data visualization, working with University of Chicago Police Department incident reports to create a crime map of campus and the surrounding Hyde Park neighborhood. But in this case, the data was not available in an easily usable format, requiring Huynh and Xu to scrape UCPD incident reports to generate their own crime dataset. They then divided the incidents into three categories -- violent crime, non-violent crime, and non-crime incidents -- and mapped them using knowledge acquired at the Leaflet workshop on Friday night.

A screenshot from Ben Huynh and Peter Xu's visualization of UCPD incident report data.

A screenshot from Ben Huynh and Peter Xu's visualization of UCPD incident report data.

In addition to taking home the Printrbot for best project, Huynh and Xu also won an unofficial competition as the only team to stay for the entirety of the hackathon.

“I think I spent about 26 hours at the CI,” Huynh said. “Peter got 2 or 3 hours of sleep; I didn't sleep until we decided we were done. Coding for 20 hours straight surprisingly turned out to be pretty fun. Our project wasn't excessively ambitious, and the CI workspaces were really nice and comfortable, so it made for a pretty steady, relaxed feeling while working. The copious amounts of free food definitely helped.”