By Rayid Ghani

We wanted to share an update with you from the Data Science for Social Good Fellowship team. Based on the funding situation and other projects that we’re launching and continuing, we have decided not to run a DSSG summer fellowship this summer. I know this will come as a disappointment to a lot of you and although we share that feeling, we are also excited about the new programs we’re working on (including an expanded/extended DSSG, supporting and launching other DSSG programs around the world, and more).

The first four years of DSSG were generously funded by Eric and Wendy Schmidt through the Schmidt Family Foundation. After the end of that funding, we had to look for additional funding sources to continue the program. Frankly, we were a little late getting started with the fundraising efforts. Earlier this month, we were at a point where we had to decide whether to:
1) run a smaller version of DSSG with the limited funding and resources we had, or
2) skip one year of the program as it has been run and focus on other programs we are launching or expanding.

After a lot of deliberation, we decided to do #2. Here were our reasons:

Although we love DSSG and have had a great time doing it the last 4 summers, it takes up all of our time and resources and doesn’t allow us to do much of anything else. If we had enough funding early enough in the cycle, we would’ve been able to build a team focused on DSSG, allowing the rest of the Center for Data Science & Public Policy to continue the projects we already have going and to make sure we honor our commitments to our other partners.

What we’ve figured out how to do in the Summer Fellowship:  Training, Prototyping, Creating use cases and case studies

Based on our experience the first four years, we realize that we’ve got a pretty good idea now on how to train aspiring data scientists who come from different disciplines to work in teams on social good problems. The program, we think, has gotten better each year as we learned from some of our mistakes and fine-tuned the training content, projects, and structure. We’ve been able to bring in ~40 fellows every year, start 12 new projects every summer, and in addition to training fellows, we’ve produced working prototypes in 12 weeks that show the potential of data science for governments and non-profits. These projects are not only directly useful for our project partners, but also serve as examples of the use of data-driven approaches for other government agencies and non-profits.

What we haven’t done (a lot of) in the Summer Fellowship: Deploying our solutions, Increasing internal data analytics capacity of governments and non-profits, Training students who want to come in with minimal technical skills

Because of the structure and duration of the summer program, we haven’t been able to successfully transition many of our projects to our project partners so that they can be deployed and operationalized. Doing that well has, unsurprisingly, required us to continue the project after the summer, which we have only been able to do in a handful of cases, given our limited (funding) resources.

We also haven’t been able to focus (in the summer) on increasing the internal analytics capabilities of governments and non-profits. We are running other programs focused on that (such as the Applied Data Analytics program launched last week jointly with NYU and University of Maryland) as well as trainings we have been doing as part of the Civic Analytics Network.

One decision that is painful to make every year is to not admit fellows to DSSG who apply with minimal technical skills and lack programming experience. Every year we review a lot of extremely smart and passionate students working in the social sciences or public policy who want to come to DSSG. While we desperately want to help them (and admire their passion and commitment to solving important social problems), we don’t think DSSG can benefit them in the same way as those coming in with more advanced technical and programming skills. We often encourage them to pick up those skills somewhere else and then apply to DSSG again.

What are we going to do instead of DSSG this summer? Create a worldwide DSSG Network, Year-round expanded DSSG, Two-week summer bootcamp

DSSG Network and DSSG Europe: We’ve realized that after four summers of DSSG, we’ve created a template for this fellowship program that is being replicated by other universities around the world (GeorgiaTech, UW Seattle, and IBM programs are three examples). We want to help support those programs and are now helping launch and run a DSSG Europe this summer that will take place in Portugal. The first iteration of DSSG Europe will be a much smaller program than DSSG Chicago but will be structured similarly. We hope to help it grow across Europe as well as other continents as part of a worldwide DSSG Network. If you’re interested in being part of the Data Science for Social Good Network, please contact us.

Expanded Year-round DSSG: We’re putting together a plan to start raising funds for a year-round expanded DSSG program that will tackle some of the gaps we’ve identified, such as the difficulty in transitioning the project back to the project partners at the end of the summer and building their internal capacity. We also want to help other universities interested in getting involved in data science for social good projects create collaborations with their local governments and non-profits. Here’s the program we’re thinking about currently:

  • DSSG activities will kick off in the fall with some training workshops for project partners as well as universities who are interested in getting involved, mostly focusing on project scoping.
  • Once the project partners submit project proposals to be considered, we’ll involve other interested universities in the project selection and scoping process, allowing them to learn how to identify and scope these projects as well as how to build successful collaborations with local governments and non-profits.
  • Fellow applications will be for two types of fellows: summer and year-long.
  • DSSG summer program will kick off in the summer with ~12 projects and ~40 fellows with two changes:
    1. We will ask project partners to embed one person from their organization in the summer team so they can go through the same training and experience as the fellows.
    2. We will create teams so that every team has at least one person who is accepted as a year-long fellow.
  • Once the summer is over, a subset of the projects will be continued with the year-long fellow deployed to the project partner for the next 6-9 months. This will create a cohort of project partners and deployed year-long fellows that will help transition the project back to the partner, allow them to validate and deploy the system, and build their internal capacity. We will continue to support the fellows and partners through convenings, workshops, and other trainings.

This is still very early in the planning process so things will change as we talk more about this idea and receive feedback from others.

Two-week Summer Data Science Bootcamp for Social Scientists and Policy researchers and students: To support and train social scientists and policy students and researchers, we are coming up with a plan to run a two-week data science bootcamp. We want to take our experiences running DSSG and other training programs and provide training to get students up to speed in Python, databases, machine learning, and other computational data analysis methods and tools that we think are necessary in doing large scale social science and public policy work. More details on this program will be coming soon.

What else are we up to? 

While we’re sad that we won’t be running DSSG this summer, we are excited about the new things on the horizon and about the projects we are currently working on at the Center for Data Science and Public Policy at the University of Chicago, including our work in:

  • Criminal Justice: identifying individuals who may be at risk of going to jail and are in need of health and social services – in collaboration with Johnson County (KS) and Salt Lake County (UT).
  • Public Safety: building an early intervention system to prevent adverse police interactions with the public with several police departments and deploying it in Charlotte Mecklenburg and Metro Nashville Police Departments.
  • Workforce Development: building a workforce data platform that employers, government and non-profit training agencies, and researchers can use to better understand real-time supply and demand of skills at a local level.
  • Public Health: deploying our preventative lead poisoning inspection tool in the City of Chicago and starting new efforts around HIV Prevention in Chicago.

Across all of these projects and more, we have been working on developing methods for auditing/detecting (and correcting) bias and discrimination in data-driven systems. We believe that there is a need not only for highlighting the potential risk but also providing guidelines that these systems can be evaluated against, and tools that can audit and suggest ways of correcting the bias and discrimination. As we work across criminal justice, public safety, public health, education, social services, workforce development, and economic development, we hope to be able to study these issues as they affect people and agencies in different settings and believe that we can help develop guidelines, methods, and tools that can be used by a variety of government agencies and non-profits.

We will still be hiring full-time staff and summer interns to work on these projects this summer so if you’re interested in those positions, let us know.

If you’re interested in helping organize, be part of, attend, or fund any of the programs mentioned above, contact us.