Brandon Rohrer, Data Scientist at Facebook, on Best Practices in Data Science for Good at ODSC East 2018
ConferencesData Science for GoodAlgorithms|Data analysis|Facebook|Social Network AnalysisBig Data|Data analysis|ODSC East 2018|Social Network AnalysisSocial Impactposted by Alex Amari August 6, 2018 Alex Amari
Key Takeaways about using Data Science for Good:
- Facebook is working hard on ways to use data science for good to solve big problems.
- Using data generated by the ‘safety check’ and ‘community help’ functions on the platform, Brandon Rohrer and his team have helped create disaster maps to aid first responders in the wake of natural disasters
- The team employs a variety of methods such as de-identification of individual data points, as well as temporal and spatial aggregation and smoothing to ensure that private data remains private.
Brandon Rohrer has spent much of his career as a data scientist at Facebook thinking about ways that data can help solve big problems. In his talk at ODSC East 2018, Rohrer showed how Facebook has used data to assist in humanitarian relief efforts after major natural disasters.
Slide copyright Brandon Rohrer, ODSC East 2018
Most Facebook users are familiar with the features ‘safety check,’ which allows users to mark themselves safe in the event of a natural disaster or other emergencies, and ‘community help,’ which aims to make it easier for people affected by disasters to find each other locally and/or provide or receive help. But beyond the practical uses for marking oneself safe or connecting with others in the event of an emergency, Facebook has sought ways to use the data generated by these functions to generate vital information for first responders and humanitarian organizations.
As Rohrer explained, the data from safety checks and community help, when aggregated, can be extremely helpful to organizations like the Red Cross and the Bill & Melinda Gates Foundation. The location data can help rescuers and aid organizations know when individuals are leaving and where they’re going. For example, when a wildfire ravaged hundreds of homes in Laguna Verde, Chile, Facebook data generated a detailed picture of evacuation patterns that benefited first responders:
Slide copyright Brandon Rohrer, ODSC East 2018
In another case, safety check data was used for damage assessment and emergency aid provision following the 2016 Kaikoura earthquake, a magnitude 7.8 earthquake in the South Island of New Zealand.
Slide copyright Brandon Rohrer, ODSC East 2018
Facebook is increasingly concerned with ensuring that private data remains private and that personally identifiable information (PII) doesn’t fall into the wrong hands. Facebook has taken several steps to mitigate these risks. First, deidentification of the data takes place such that all PII is removed from individual data points. Next, techniques for temporal and spatial aggregation and smoothing are employed to conceal individual data points. Finally, Facebook only distributes the data to trusted organizations like the Red Cross and the World Bank.

Alex Amari
I’m a graduate student at Oxford University pursuing an MSc in Social Data Science with the ultimate goal of working in tech entrepreneurship.
Top Deep Learning Sessions Coming to ODSC West
West 2023Conferencesposted by ODSC Team Oct 4, 2023
Anomaly Detection for CRM Data: A Step-By-Step Guide
West 2023Modelingposted by ODSC Community Oct 3, 2023
How to Make Your Own ODSC West Schedule
Conferencesposted by ODSC Team Oct 3, 2023