I’m an Associate Professor of Geography and Information Science at the University of Colorado and a Data Scientist at Apple. My expertise is at the intersection of maps, statistics, machine learning, and the social sciences. Our recently published book on Urban Analytics tries bring together these diverse fields.
I have received the Breheny Prize for the best paper in Urban Analytics and City Science, I won a kaggle.com data science competition, and was awarded a distinguished scholar award in Urban Planning from the American Association of Geographers. The journal Science profiled me as an archetype of a new generation of data-centric geographers. My publications have appeared in a diverse set of journals including PNAS, PlosOne, Demography, Annals of the Association of American Geographers, and the International Journal of GIS. Outside of academia my professional experience has ranged from the hyper digital world of Data Science and Software Engineering for a large tech company in Silicon Valley to the insanely analog practice of being the sole proprietor of an antiquarian bookshop in Manhattan.
As a geographer I’m interested in how we map the world. I know “mapping the world” seems old school, really old school, like an 18th or 19th century problem so some explanation is necessary. As a researcher I develop better ways to digitally represent social space and statistically reason with those representations. That sounds super academic (and it is), so let me provide some practical examples.
The federal government allocates $400 Billion to communities in the United States each year using a map from the US Census Bureau, this map describes demographic and economic conditions in every state, county, city, and neighborhood in the US. In theory, this data allows the federal government to target aid to communities in need. These Census maps show things like the percent of the population in poverty or the number of school age children. These are not maps of rivers, roads, and mountains – but they’re maps.
What if that poverty map is wrong? How would we know if it was wrong? People in need might not get aid and people who don’t need help might get it. Maps that represent things that are hard to see, like social space, are difficult to create and evaluate. In my academic work I study these maps, develop software tools to make these maps better and create statistical (and ML) techniques to analyze such maps.
Maps describing social space are used widely in the sciences, business, and governance. I am especially interested in the concept of “neighborhoods”- I’ve developed methods for measuring neighborhoods using individual-level data and have studied the relationship between neighborhoods and well-being. I served on the Department of Interior Hurricane Sandy Strategic Science Group and in a recent issue of PNAS I developed a new way to measure neighborhood-level social vulnerability to natural hazards.
Prior to joining the faculty at Colorado I was the Associate Director of the S4 Initiative at Brown University. Before Brown I was at the Columbia University Graduate School of Architecture, Planning and Preservation, first as a masters student and later as an Adj. Assistant Professor and Associate Research Scientist. Once upon a time I was the owner of an antiquarian bookshop in the Flatiron Building in New York City. I’m also a fisherman, a bird-watcher, a really good snowboarder, and a bike geek. I program in R, Python, and Scala (badly) and whenever possible I work to make my research reproducible by publishing data and code.
Note: During 2015-16 academic year I am on leave from the University of Colorado working at Apple. From 2017 on I will be dividing my time between Apple and the University of Colorado.