Calling all curious data scientists and social scientists

Data Science for Social Good ATL is your opportunity to make a real difference in civic and social issues in Atlanta. In this summer program, you’ll help nonprofits and city organizations solve their most pressing problems. You’ll use a combination of data science and social science skills to build tools your partner organizations can use right away.

You in? Here’s what this might look like.

When you start the project, your first step is to find the problem

Yes, you will start with a project description. But it’s often incomplete. Humans don’t often know what they want: as Henry Ford said, “If I had asked people what they wanted, they would have said, ‘A faster horse.’”

Your job, then, is to find out what will really help.

To do that, you spend the first two weeks in your partner organization. Get to know the team. Who are they? What do they do on a daily basis? Where are they frustrated, and where do things work well?

Inevitably, people will tell you that they really want you to do X. You’ll say: “Great! What would X allow you to do?”

For example: “We really want you to improve our 911 response times.”

You: “Great! What would improving your 911 response times allow you to do?”

Them: “Save more lives.”

Boom. You’ve found the real problem. No one cares about response times in isolation–it’s that response times allow emergency responders to save more lives.

Keep asking, “What would X allow you to do?” until you find the key problem.

Then ask: “How would you know if you’ve accomplished Y?”

In this example, you might hear: “When the number of lives saved increases by 10%.”

Great. Now you have a problem and a definition of success.

Next, you’ll develop a solution

Based on everything you’ve learned so far, how do you plan to approach the problem?

Sketch out a few different possibilities. Then run small tests on each one. How can you quickly find out which ones to use and which ones to toss?

Once you’ve got an idea of which approach to stick with, reach out to your Georgia Tech mentors. Also ask for coffee meetings with local data scientists. Share the problem, the definition of success, what you’ve tried, and what you plan to do. Then ask: “If you were in my shoes, what would you do differently?”

Revise your approach based on the feedback you get. Repeat this cycle until you’re sure you’ve got it.

Then you’ll field test your solution

Here’s the fun part: share your solution with your partner organization. Show your clients how to use your tool. Then let them run with it for a week.

Now is the time to observe how your client uses your tool. Ask the same kinds of questions you asked when you were finding the problem. Then record your own notes:

What works beautifully? Where are people frustrated? What are they trying to do that you hadn’t anticipated?

Use your field test to improve your solution. Then repeat the experiment: share the next version with your partner organization. Keep it up until you’ve reached the definition of success you laid out at the beginning of the project.

You want to do this because you’re curious. You love the idea of crafting a new technical solution. The thought of hanging out in partner organizations to see how they work gets you pumped up and ready to go. You know you’ll feel amazing when you deliver something that actually helps those partner organizations do their work better.

You also have a solid foundation in technical skills. You’ve taken at least one class in statistics, machine learning, or data mining. When presented with a problem and the data available, you can sketch out a few different approaches to the problem. You might even have some basic programming skills.

Just as important, you know how to communicate technical results. You’re ready to tell your partner organization up front what they should do, then save all the statistical analysis on why they should do it for the Q&A period. You write well, and you love presenting recommendations to a non-academic audience.

You’re thrilled about working with top-notch people whose skills complement your own. Maybe you’re a programmer, and you’re excited to learn about policy analysis. Or maybe you’re a social scientist, and you’re excited to bolster your technical skills.

Does this sound like you? Are you ready to solve real policy problems with data?

If you are a student, apply now.

If you want to help as a mentor or partner organization, drop us a line.

Brainstorming session

Earlier this week, we held a brainstorming session at InvestAtlanta designed to generate civic ideas. Here are a few of our favorites from that session.

1. Identify the best neighborhoods for investment based on City and demographic data. What factors predict the return on investment in a neighborhood? How can City officials use data to create a bigger impact with the same amount of money?

2. Improve the 9-1-1 emergency response system using call audio, location and response information. If data scientists could automate the transcription of calls, what patterns would emerge to help guide dispatch calls? Could data analysis tell us if calls following certain patterns (location, time) get a faster response than others?

3. Measure the structure, network and impact of the startup ecosystem in Atlanta. What does the network of startups and their related organizations in Atlanta look like? How can we measure the impact of the startup ecosystem on the city as a whole? Which parts of the ecosystem need more attention to make Atlanta an even more awesome place for startups?

4. Prioritize city infrastructure improvements using data gathered by the upcoming 3-1-1 portal.  How can the limited resources available for infrastructure improvements be used for maximum impact? For example, how can city officials know which potholes are the most dangerous or annoying to citizens, based on 3-1-1 data?