Data Science for Social Good Atlanta (DSSG-ATL) students and mentors concluded another year of solving problems for the city of Atlanta and beyond. The annual student showcase took place July 24 at Ponce City Market, with nearly 75 people in attendance including data scientists, local companies, non-profit agencies, and organizations.
“This fourth summer of the program has been a huge success,” said Ellen Zegura, Stephen Fleming Chair of Telecommunications in the Georgia Tech School of Computer Science, and the one who began Atlanta’s DSSG program. She kicked off the final event, which shared innovative data-driven approaches and results for five projects in the areas of housing justice, food security, crowd-sourced environmental monitoring, flood prediction, and building energy consumption.
DSSG-ATL is an intensive, ten-week paid internship experience that blends data science and technology design. Students are placed on multi-disciplinary teams and matched with a supervising professor to address real-world problems with partners in the city of Atlanta and local non-profit organizations. DSSG teams blend expertise on technical topics, public policy, social issues, and education for a truly interdisciplinary approach.
“DSSG connects the classroom with real problems of deep community relevance. We hope this will inspire them to pursue their technical education further and to be engaged global citizens that use their education for societal impact,” said Bistra Dilkina, School of Computational Science and Engineering assistant professor and DSSG co-director.
The program simultaneously addresses several emerging concerns: finding innovative solutions that serve immediate needs, getting the experience necessary for developing the nation’s data science workforce, and helping students communicate effectively by working with actual clients and team members.
“It is a unique opportunity to make a real impact in the Atlanta community. Students get the ability to understand how to communicate data problems with organizations that are not coming at this with a high level of technical background,” said DSSG-ATL Co-Director Chris LeDantec.
The highly competitive internship selected 17 interns from around the country from a pool of over a hundred applicants with backgrounds in computer science, statistics, digital media, public policy, civil engineering, industrial engineering, and urban planning.
This year, DSSG received sponsorship from the National Science Foundation and the South Big Data Innovation Hub at the program level.
During the first week, students participated in hands-on tutorials in data analytics, web stack development, spatial information, and geographic information systems. In the remaining weeks, students worked with government and nonprofit partners to design data-driven solutions. They also attended weekly seminars with invited speakers from academia and industry on technical topics and prominent examples of using data for good.
By working with real data and organizations, the students develop useful tools. “Some of these apps will be used by the actual agencies, so that means they’re really influencing people’s lives,” said South Big Data Hub Co-Executive Director Renata Rawlings-Goss.
They also learn critical skills such as stakeholder engagement, data acquisition and processing, data analysis and visualization, machine learning for predictive modeling, writing, and communicating results to nontechnical audiences.
During the summer of 2017, students worked on projects targeting five social problems.
Food for Thought: Analyzing Public Opinion on the Supplemental Nutrition Assistance Program
Interns: Miriam Chappelka, Jihwan Oh, Dorris Scott, Mizzani Walker-Holmes
Georgia Tech Program Mentor: Carl DiSalvo, associate professor in the Ivan Allen College of Liberal Arts
Community Partner: Atlanta Community Food Bank
A major tool in the fight against hunger is the Supplemental Nutrition Assistance Program (SNAP), commonly known as food stamps. Understanding the news cycles, geography, and changing attitudes of the population toward SNAP is useful for shaping policy. Teams used data from Census data, voting records, news articles, Twitter, and Facebook to categorize sentiments and conduct text mining. The team created three products: an InfoMap to view the geography and some correlates of SNAP-related attitudes, a voting records database to prepare for meetings with policymakers, and a Visualization of Sentiment Analysis to assess sentiment about SNAP during major events.
Cycle Atlanta: Seeing Like a Bike
Interns: Javier Argota, Myeong Lee, Noel Mannariat, Erica Pantoja
Georgia Tech Program Mentor: Chris LeDantec, associate professor in the Ivan Allen College of Liberal Arts
Community Partner: City of Atlanta
Not all bikers feel comfortable cycling throughout Atlanta, especially during rush hour. The team designed multi-sensor boxes to generate data on factors such as traffic speed, traffic volume, percent of heavy vehicles, proximity to traffic, lane configurations, pavement conditions, and many others that account for riders’ stress. Their analyses involved the creation of a Level of Traffic Stress (LTS) model and the use of two machine learning algorithms. The goal is to provide reliable data-driven evidence that helps policy makers refine cycling infrastructure and environmental conditions for cyclists of all comfort levels.
Atlanta Housing Justice: The Anti-Displacement Tax Fund
Interns: Jeremy Auerbach, Hayley Barton, Takeria Blunt, Vishwamitra Chaganti, Bhavya Ghai
Georgia Tech Program Mentors: Christopher Blackburn, Ph.D. candidate in the Ivan Allen College of Liberal Arts; Amanda Meng, research associate in the College of Computing; Ellen Zegura, professor in the College of Computing
Community Partner: Atlanta Legal Aid Society
Urban revitalization on Atlanta’s Westside, including construction of the Beltline and a new stadium, has caused increased property values and fears that residents will be displaced. An anti-displacement tax fund exists to offset the increase in property taxes for eligible residents. However, stakeholders need to understand program cost, and the number of people impacted. This project determined the number of eligible homeowners using location, owner-occupancy, property lien data, and an income model based on home characteristics. It forecasted home appreciation and property tax increases and created an interactive web app for community members to view eligibility and estimated property tax increases. The team’s projections show a higher cost and greater number of eligible households than those previously released, highlighting the value of data science techniques and community participation.
Building Energy Analytics
Interns: Keyan Halperin, Lingzi Hong, Brendon Machado, and Ricardo Macias
Georgia Tech Program Mentor: Bistra Dilkina, assistant professor in the College of Computing
Community Partner: Georgia Tech Facilities Management
All Georgia Tech buildings have sensors that assess how much energy is being used every 15 minutes over the past few years. Despite the ongoing data collection, it has largely remained unexamined. The team used the energy data from four buildings, along with local weather and building occupancy to model energy usage at Georgia Tech. They determined factors related to energy usage and predicted how it changes with related events. Using six models, they predicted energy usage, common denominators, and unique aspects of a particular building that accounted for the most energy consumption. For example, the number of classes was by far the most important predictor of energy consumption in all buildings. But temperature impacted the buildings differently. Future work can extend models throughout campus and identify inefficient buildings for upgrades.
Predicting and Alleviating Road Flooding in Senegal
Interns: Keyan Halperin, Lingzi Hong, Brendon Machado, and Ricardo Macias
Georgia Tech Program Mentor: Bistra Dilkina, assistant professor in the College of Computing at Georgia Tech
Community Partner: United Nations Global Pulse
Climate change has the potential to raise the risk of flood for coastal countries, impacting the living environment and threatening the success of crucial city development. The team created models to determine which populations or regions are most vulnerable to disruptions to movement caused by flooding, and which roads should be targeted for mitigation. First, they modeled the flooding risk of each road segment based on historical weather and topographical data. They then modeled the volume of traffic of each road to quantify how necessary it is for accessibility between different parts of the country. Phone records indicating when a customer used each tower location helped to establish road importance, while traffic flow was assigned based on the data volume. Putting it all together, the team determined the consequences and mitigation strategies for traffic flows redirected due to flooded, inaccessible roads.
Several DSSG-ATL projects from previous years have received recognition:
- Predicting Atlanta’s Fire Risk
- Optimizing Atlanta’s 911 System with Data Science
- Preserving the City of Trees with a Planting Visualization Tool
- Resettling Georgia’s Refugees in New Communities
DSSG-ATL is part of a broader community of Data Science for Social Good programs. This includes the original DSSG program organized by the University of Chicago, as well as newer programs at the University of Washington, the University of Massachusetts Amherst, and most recently at the IBM T.J. Watson Research Center. Participants become part of a network of students, mentors, professors, and projects taking place around the U.S.