These days, data is power. However, there is a disconnect between the people who collect information and those who can turn it into something meaningful. This is increasingly becoming a challenge for entities whose work may affect scores of people, such as cities, hospital systems and non-profit organizations.
Over the past three months, the first Eric and Wendy Schmidt Data Science for Social Good Summer Fellowship put 36 students to work attempting to solve these problems. On Tuesday evening at the University of Chicago’s Gleacher Center, the 12 teams presented their projects, which ranged from transportation solutions at the city level to mining tweets for crisis management across the globe.
Rayid Ghani, Obama for America’s former Chief Data Scientist, directed the fellowship for the University of Chicago, providing guidance and mentorship to the students throughout the summer.
“The goal was to train people...who not only have the right skills, but who care about making a social impact,” Ghani said in his opening remarks. Based on the projects that came to be in just three months—some of which are already functional—it seems that goal may have been achieved.
Here are the projects the teams presented at the first ever Data Slam, many of which will be released as open-source:
Problem: When disasters occur, social media provides a bevy of data, including where and how severe the damage is.
Solution: A tool that extracts data relevant to first responders from social media posts, providing an accurate account of each disaster.
Problem: Code Blues, or alerts when a patient goes into cardiac arrest, strain a hospital’s time and resources and don’t always result in saved lives.
Solution: A predictive tool that helps healthcare workers determine whether a patient is likely to enter cardiac arrest, allowing them to act preventatively.
Problem: City jails are overcrowded, and administrators don’t know which inmates are good candidates for ankle bracelet surveillance.
Solution: An analysis of the interaction of violence and incarceration that shows that inmates convicted of drug-related crimes do not correlate to spikes in crime following their release.
Problem: Adolescents from certain racial and socioeconomic backgrounds often fail to attend institutions that match their academic level, a problem known as “undermatching.”
Solution: Match.edu helps identify high school students at risk of undermatching and provides them guidance and resources to help them apply to appropriate academic institutions.
Problem: Cities collect massive amounts of data, but don’t know how to sort through it to improve municipal services.
Solution: The City of Big Data analyzes garbage collection routes, helping the Department of Streets and Sanitation to make pickups more efficient, and the correlation between streetlight outages and crimes in various areas, which the City can use to figure out which lights to repair first.
Problem: Across the world, individuals witness crises but don’t know how to report them.
Solution: A system to retrieve sift through and distribute valid crisis information, using a computer to enable action from other parties such as NGOs and journalists.
Problem: A national program needed to find out what impact their home visitation nursing program has on first-time mothers.
Solution: Data analysis revealed that mothers who participated in the program were more likely to secure adequate immunization for their infants the general population.
Problem: Homes across Cook County are under foreclosure and abandoned, but the Land Bank doesn’t know which are worth taking over and what to do with them.
Solution: Built on the Google Maps API, a program that identifies foreclosed homes and provides relevant details to help the Land Bank make sound decisions.
Problem: Bike sharing programs struggle to manually empty and fill stations to during peak usage hours and need to know how to balance where the bikes end up.
Solution: A tool that predicts which Divvy stations will be full or empty at various times, allowing the company to more efficiently move bikes around.
Problem: Commercial buildings gobble up energy but don’t always know how to decrease consumption.
Solution: A model that analyzes hourly consumption data and determines how buildings perform against other similar buildings, pointing out areas that can be improved.
Problem: The CTA is overcrowded, and officials cannot predict how rescheduling will affect this problem.
Solution: A proactive tool that allows the CTA to simulate graphs that represent the way crowding can be reduced by different schedule proposals.
Problem: NGOs in the social sector need to build relationships and seek funding, but they don’t know where to turn or who to approach.
Solution: A cloud-like network visualization that mined online text to identify relationships with peer non-profits and their funding sources.
Photo by Jason Smith/University of Chicago.