Duke, NYU, & other schools join with Bayes Impact to let students solve big problems with big data

Bayes Impact, the big data-focused nonprofit organization, is launching Bayes Impact University Program at five pilot universities this fall. The idea is to give master’s students opportunities to solve real-world problems using big data.

The University of California, Berkeley, Duke University; New York University; Northwestern University; and the University of San Francisco are participating.

“We have actually more schools that were interested than we can handle,” said Andrew Jiang, Bayes Impact’s executive director, said in an interview with VentureBeat. “But [we] decided to start with fewer schools to make sure we can be successful in the program.”

Bayes Impact targets master’s students pursuing degrees in data science, statistics, computer science, and related fields.

The program is a novel way to match great talent with practical challenges, which makes sense given that the demand for data scientists exceeds the supply, and given that as training programs proliferate, more realistic training can amount to a differentiator. The Science to Data Science program in London, for instance, also focuses on solving practical problems as it trains up Ph.Ds and scientists. But Bayes Impact’s focus on social problems gives the program a special twist.

At least on the university level, participating schools should be able to stand out with the addition of this program.

“Generally what happens with most master programs is, in the last year, they have capstone programs, where they apply their skills to a real project, or a manufactured project,” said Jiang. “What we provide them is instead of doing a project for, let’s say, Yelp, or Twitter, they are doing a project for Red Cross, or a project with another nonprofit or civic organization.”

Students in the program will work in teams of two to four on a single project for six to 10 months. Past projects range from “developing fraud detection and credit risk models for micro-finance nonprofits” to “identifying predictive indicators and biomarkers for Parkinson’s disease.”

Everything is on site for the program. Bayes Impact provides students with nonprofit or civic clients that are local organizations or global organizations that have a local office students can visit.

“They get the full experience what it is like directly making a difference on an organization,” said Jiang. “It’s hard to do it over the phone or over email. It’s really great when the student can go to a homeless shelter and see the impact of their work.”

Bayes Impact also provides students with strong mentorship from data science industry veterans and faculty mentors within the universities. For example, Bayes Impact looks for data scientists from the Research Triangle in North Carolina to mentor students at Duke University.

And these two groups of mentors complement each other, from Jiang’s perspective. Industry people provide students with practical skills and solutions. Professors bring domain expertise in addition to their data science experience, especially when it comes to health care and education.

Universities are interested in this program with Bayes Impact for several reasons. “For universities teaching data science, a real practical project is the most effective way for them to teach, [students] actually working on a real data science problem, with messy data, with an actual client to work with,” said Jiang.

It’s also the case that “most universities want to support social good efforts, they want to teach their students what else they can do besides the traditional route of data science of financial technology and ad tech.”

Bayes Impact is backed by Y Combinator and will be presenting at YC’s demo day on August 19th.

It launched its Fall 2014 fellowship last month. More than 200 people applied for eight openings, Jiang said. Their backgrounds range from senior data scientists at Google to computational physicists at Harvard and Stanford, he said.