Rayid Ghani: From the Campaign to Computation

17
Apr
2013

If you received a surprisingly personalized e-mail or Facebook message from the Obama 2012 campaign, it was likely the product of the campaign's groundbreaking analytics tools. As chief scientist of that acclaimed team, Rayid Ghani helped bring the computational techniques of data-mining, machine learning and network analysis to the political world, helping the re-election campaign raise funds and get out the vote in powerful new ways. Now that Barack Obama is back in the White House, we are pleased to announce that Ghani is joining the University of Chicago and the Computation Institute. Here, he will shift his attention and expertise to even bigger goals: using data and computation to address complex social problems in education, healthcare, public safety, transportation and energy.

Though he only started on April 1st, Ghani already has a full plate, including a position as Chief Data Scientist at the Urban Center for Computation and Data and a role developing a new data-driven curriculum at the Harris School for Public Policy. But Ghani's most immediate project is The Eric and Wendy Schmidt Data Science for Social Good Fellowship, which hopes to train and seed a new community of scientists interested in applying statistics, data and programming skills to society's greatest challenges. We spoke to Ghani about his time with the campaign and plans for the future.

{C}

Q: So what brought you to the University of Chicago and the Computation Institute?

Ghani: The reason I got involved with the campaign was that I was looking to combine the things that I care about with the things that I'm good at. I was good at machine learning and data mining research and I cared about making a social impact in the world. The campaign was the beginning of that, but not a long-term plan. After the campaign, I was even more enthusiastic  - if we could do all that we did in a year and a half, there's certainly a lot more we can do if there is a more focused effort that can last.

I started talking to non-profits, foundations and companies trying to figure out the best place to do data analytics work that can benefit society in general and solve large social problems. It seemed like there was already a lot of this type of work already going at the University, at Harris, at the CI, where people were looking at similar problems.

What brought me here was a convergence of two goals: I want to help create people who not only have the skills to solve large-scale data and analytics problems but who are also passionate and excited about making a social impact. I also wanted to work on those problems at the same time, bringing people together from computer science, statistics, policy and social sciences. Being here helps me bring a lot of these people together and accomplish a lot more than I can do just by myself.

Q: Where do you think there is potential for this kind of work to make difference?

Ghani: I'm interested in large social problems that impact individuals. People have focused a lot recently on collecting data on individuals; schools have been collecting data on students, healthcare policy changes are forcing data collection for compliance reasons. A natural next step is to use it to improve outcomes for all these different areas. In the case of education, how can we look at individual student behavior and their environment and their performance and predict who is at risk of dropping out of high school or not applying to college.  Same for child obesity – can we predict who is at risk?

Once you’re able to do that and identify at-risk individuals, you then design intervention to improve these outcomes. It's useful to know if people are at risk, but if you can't do anything about it, that doesn't really help anyone. There are different kinds of interventions that work for different sets of people, and then the questions are which intervention is the right one, how do you deliver it, and who delivers it? We're at the point where a lot of people have spent a lot of time doing predictions and now is the time to take the next step.

Q: How did your time with the Obama campaign help you define these goals?

Ghani: The campaign was helpful in seeing that it can be done at scale. If you put the right people together and build the right tools and motivate people the right way, you can mobilize them for a specific cause. It's a really good case study of what is possible for very large social problems. Yet is a presidential campaign so unique that it will only work there? That's an open question, and something we want to explore.

The campaign helped formulate an approach that can be reused across a lot of areas, where you've got data about people coming from all sorts of different places, you have predictions about people's behavior, who you intervene with, and which interventions work better. These ideas have existed before, but before the campaign, there wasn’t much work done at such a large scale to show the effectiveness of this kind of approach. It really helped show people that it can be done.

Q: What is the idea behind the Data Science for Social Good fellowship?

Ghani: The fellowship is exciting for a few different reasons. The kind of students we tried to recruit are those with strong computational and/or quantitative backgrounds, who are interested in working on large social problems, but haven't really had exposure to them or the opportunity, and don't know where to go or what to do. The typical companies that go recruit them are tech and web companies, and that's where they end up going because that's what they know, and everybody else they know is there, and that's what they hear about at conferences. So, the motivation for this fellowship is to train these students in solving problems that have large social impact of some sort.

We want to bring these students together with mentors and problems and organizations so they can not only solve important problems this summer, but start thinking about them more broadly. That's the goal: to solve some concrete problems in the summer but also, long-term, create a culture where a lot of these data scientists are thinking about social problems.

Written By: 
Research Tags: 
Tags: