February 16, 2017

The value of diversity in data science research

By: Aude Hofleitner

Leaning in
This week, we hosted over 35 of the top PhD candidates in the US at our Facebook Menlo Park campus for a Women in Research Lean In event. As part of the event, I joined 5 of my colleagues for a “Women in Research” panel to share some of my experiences and insights working in technology, and am sharing those more broadly here.

I am a Data Scientist on the Facebook Core Data Science team working on understanding the connection between people’s online and offline identities. The goal of my research is to help inform product teams how to deliver more personal and valuable experiences to people. Experiences that build meaningful connections with the people and places they interact with around the world. And, I get to work on really amazing things.

As a Data Scientist, I have the opportunity to apply my expertise across a range of Facebook products. Here are some examples of my work:

Understanding barriers to Internet access in low coverage areas
My interest and expertise in spatial data analysis led to a collaboration with the Internet.org team on a project where I inferred the coverage and quality of existing Internet connectivity in low-bandwidth areas. The goal of the work was to better understand some key barriers to Internet access, and help the team devise and adapt new technical solutions to people get online.

Connecting in times of crisis
It is incredibly gratifying to see the importance that Facebook Apps play in bringing people together around the world, and sometimes in very trying situations. After an earthquake occurred in Napa, CA, I noticed and wrote about how people in the affected area had significantly increased their use of the platform to communicate with friends and share updates. This is consistent with what we have seen in other times of crisis, where connecting is more important than ever.

As social media has evolved, people have turned to it as an efficient way to deliver one-to-many communications. This makes a platform like Facebook a natural place to share updates in times of crisis. At the time of the Napa earthquake, a team was already building the Facebook Safety Check, and after reviewing our findings, we were able to quickly transfer the analysis insights into the design of the feature.

During the refugee crisis last year, we saw an increased number of people show their support on Facebook. We developed a machine learning algorithm to classify pages and groups as being supportive of the Syrian refugee crisis and were able to show that nearly 8 million people around the world used more than 35,000 Facebook Groups and Pages to support the Syrian refugees. We also found that four out of five people on Facebook are connected to someone directly affected by the Syrian refugee crisis through a friend of a friend. This demonstrates the impact of connecting in a global world. This research was featured in the “You’re not alone” video from Mercy Corps, International Rescue Committee.

A Data Scientist’s toolset
As a Data Scientist, I frame each project using data exploration and visualization techniques. This also helps me set the goals for the project to keep it focused. As I work, I develop algorithms to blend spatial data analysis, machine learning, statistics and graph inferences. The scale of Facebook data is both a treasure trove and a challenge. I had not had access to this level of data prior to joining Facebook, and my skills in algorithm modeling and design have helped me assimilate large amounts of data to attain valuable insights.

Making immediate impact is also a benefit of my work at Facebook. I interact daily with highly cross-functional teams to evolve early research ideas into product features used by billions of people around the world. Being involved throughout the product development cycle is interesting and empowering.

My immediate team and the people I work with are gender, ethnically and geographically diverse. Working with such a diverse set of people is also essential to building products that are relevant to people around the world.

The value of diversity
I completed my undergraduate and masters studies in France in the classic “Classes Preparatoires – Grandes Ecoles” system. It was rare for someone from my high school to follow that path so I was initially intimidated. However, my high-school math teacher convinced me to give it a try and somehow it worked out better than I could have imagined. I joined Berkeley to pursue a PhD in EECS in the Controls program, working with Partial Differential Equations, optimization and statistics. Given my training in math and applied math, I had not considered a career in industry, and even less in a “software engineering” company. Regardless, I decided to intern at Facebook during my PhD to see how my skills might be applied. What surprised me most was how much the diversity of my colleagues backgrounds were embraced and what each person’s unique perspective and knowledge brought to the table. From theoretical computer scientists, physicists, political scientists, sociologists, to systems engineers, the sum of our backgrounds and insights give life to better experiences for the people who use our products.

At both Berkeley while pursuing my PhD and now at Facebook, I have been involved in encouraging more women to pursue STEM fields. Through WICSE (Women in Computer Science and Electrical Engineering) at Berkeley, I organized events, mentored students and participated in outreach events. At Facebook I’ve been fortunate to participate in many diversity events such as the Lean In one this week, and Stanford’s Women in Data Science (WIDS) conference a couple of weeks ago, this is a great way to offer support, advice and encourage women to discover new areas and opportunities they may not have previously been aware of.

As a French woman, with a PhD in EECS, working in a US technology company, I was supported and encouraged by many mentors along the way. From my high school math teacher, to my parents and later on my graduate school adviser and fellow students. Their encouragement and push helped me pursue and thrive on a once intimidating path. One that lead to where I am today. Had it not been for them I would have taken a very different journey, most likely less technical and far-less challenging. I hope to inspire a few more promising mathematicians to take this unconventional, highly rewarding road.