Brian's a statistician, journalist, and hacker. He lives in New York and serves as a 2013 Mozilla-Knight OpenNews fellow at the New York Times. Before that he was data scientist at the Harmony Institute. He recently graduated with a MA in Applied Statistics from Columbia University where he focused on quantitative and computational approaches to social science. On the side, he edited a book for a prominent political scientist, particiapted in hackathons, and worked on investigative news stories. Brian has managed development projects in sub-Saharan Africa, reached number one on the Hype Machine, and shared a stage with Spoon and Bob Dylan.
"If the Federal Reserve account balance falls in today’s treasury statement, does it make a sound?" Using an untapped reservoir of Federal Treasury data, this project sought to generate a dance music video to communicate the complexity of the multi-dimensional dataset. Built over a weekend at the Bicoastal Datafest, it was awarded Best in Innovation and Best in Show by a distinguished panel of judges.
An ongoing project with Michael Keller that came out of an award-winning hack at the 2012 Columbia J-School Datathon. It utilized Stop, Question, Frisk data to show a significant increase in stops of Shi'a Muslims around Mosques considered to be potential "hot spots" for terrorist activity by the NYPD.
An interactive map of mass shootings created with Michael Keller in the wake of the Aurora massacre. The map was built over a weekend from a PDF published by the Brady Campaign. Using Amazon's Mechanical Turk and some Natural Language Processing, we located news aritcles for each shooting and determined the number of people injured, killed, and whether the incident was gang related or in public. The resulting interactive included a feature for readers to share their memories of particular shootings. Since its publishing, almost 1000 people have provided their accounts of these tragic incidents.
By combining social media data, content analysis, and brain scans, this study attempted to determine the neurological patterns and emotional characteristics assoicated with activity on Twitter. Using AMC's The Walking Dead as a case study, it compared spikes in tweets during the series premiere with EEG data from 20 research subjects watching the show. The study found that moments of the show that elicited high levels of neural synchronicity were signicantly correlated with more tweets.
This project has been funded by DARPA. It was completed while working at the Harmony Institute and benefited from the help of Jacek Dmochowski and Lucas Parra of the City University of New York.
HiScore - Towards an Open Source Model of Influence
HiScore is a proposal for an open-source alternative to Klout written in code. Using data scraped from social media, web searches, and news articles, the score was tested on award-winning documentary films, showing how certain films with social messages can sustain attention beyond their theatre run.