Human-Centered, Pro Bono Data Science at Machine Eatable

| Miriam Young, Head of Communications & Culture, DataKind

Photo: Susan Sun at Machine Eatable | C/O John Paul Farmer.

Just in time for International Pro Bono Week, this past month’s Machine Eatable discussion featured two DataKind volunteers presenting on recent pro bono consulting projects. These DataCorps projects, as they’re called, typically last 3-6 months with volunteers donating a whopping 5-10 hours per week in addition to their full-time jobs. While the projects are pre-scoped by DataKind staff before teams start, there are still usually some twists and turns along the way where teams must pivot and come up with creative solutions. 

With a PhD in business economics from Harvard, DataKind volunteer Raluca Dragusanu first presented on her team’s work with a nonprofit called MicroCred. MicroCred works to make financial services available to the individuals that are underserved or unserved by the traditional financial sector, particularly the micro, small and medium entrepreneurs. In a project sponsored by IBM, her team used predictive modeling to improve their customer scoring and overall efficiency in granting loans so they can ultimately reach more people underserved and unserved by the financial sector. 

Up next, Susan Sun, a freelance data scientist working at Google, spoke about her team’s work with VOTO Mobile, one of West Africa’s fastest growing social enterprises. They aim to amplify the voice of the underheard through a mobile-based Interactive Voice Response (IVR) and survey platform that removes the barriers to insightful mobile communication between citizens worldwide and the organizations that serve them. In a project sponsored by Google, her team did statistical analysis on VOTO Mobile’s call and response data to identify the factors that result in successful completion of IVR surveys by women. This is all in an effort to combat the issue of data deserts, where certain populations tend to be underrepresented in datasets used to drive decisions around policy or other humanitarian interventions. 

Both projects had some common twists and turns including challenges with the data itself and also ensuring that the team’s work was continuously guided by the organization’s needs. As Susan described, her team had to pivot from a “hammer/nail” mentality where you want to fix everything with a machine learning model to a human-centered one where you are tailoring your work to meet the actual needs of your partner – even if it means doing something simpler or more foundational. Similarly, Raluca commented that The Data for Good movement is about more than just data science – it’s also about empowering and supporting the people that will ultimately carry the work forward.

Indeed, people are ultimately at the heart of this work so it’s important not just to select the right model or have the right data, but engage the right people to affect long-lasting change.

Tweets from this month’s #MachineEatable:

Tags: , , , , , , , , , , , , , , , ,