CGA: Illuminating Space and Time in Data Science

May 11, 2018
cga_logo_globe_400x400

The 2018 Research Conference of Harvard’s Center for Geographic Analysis examined the convergence between geographic information science and data science. " src="/profiles/openscholar/modules/contrib/wysiwyg/plugins/break/images/spacer.gif" title="<--break-->">The two-day meeting, titled Illuminating Space and Time in Data Science, brought together academic and industry experts to discuss their perspectives on a range of new technologies and techniques for analysis and interpretation of the dynamics of everyday life, environmental monitoring, and the possibilities for greater civic engagement. This new era, sometimes referred to as big data or the internet of things or pervasive computing, marks a moment in which scientists, policy makers, and consumers have access to enormous datasets to make observations about the world, to plan for a sustainable future, or to purchase the most reliable appliances.

The rapid proliferation of ‘smart’ objects has enabled a variety of sensors operating at a wide range of scales -- from the body to the planet -- resulting in unprecedented volumes of digital data. Most readers of this text have the ability to record their location at this moment, and many are probably transmitting that location now, some consciously and some unconsciously. From satellite images to social media streams, from census and parcels to records of trade, food, energy, climate, disease, crime, conflicts, etc., big data with space and time signatures are essential for understanding our world and responding to its challenges.

2018-cga-conference-1-adjusted

The ramifications of the world’s expanding data are not purely academic. “Even ‘safe’ pollution levels can be deadly,” said Prof. Francesca Dominici, Director of the Harvard Data Science Initiative, in her morning keynote address at the event. Dominici’s research team analyzed terabytes of high-spatial resolution exposure and medical data through complex computations to study air pollution impacts on human health. Within the field of data science, “Spatial is special,” noted Michael Goodchild, Emeritus Professor of Geography at the University of California, Santa Barbara, in his luncheon keynote address. Goodchild gave an overview of the landscape of geographic information science (GIScience) in which he emphasized that if data science is about prediction, then space and time enable prediction of where and when. Dominici and Goodchild headlined a day-and-a-half program with nearly 30 invited speakers, each exploring the interrelationships of space and time in data science before an audience of over 200 attendees.

The conference was co-organized by the Center for Geographic Analysis (CGA), the Harvard Data Science Initiative, and Esri, a global leading company in mapping and spatial data analytics technology. Co-sponsors included MapD Technologies, Inc. and NSF I/UCRC Spatiotemporal Innovation Center. Prof. Jason Ur, Director of the CGA, and Elizabeth Hess, Executive Director of IQSS, opened the program on Thursday and Friday respectively, both pointed out the need to explore relationship between data science and GIScience.

2018-cga-conference-3-adjusted.jpgFour technical workshops on Thursday highlighted ongoing research and applications in spatiotemporal analytics from the co-organizers and co-sponsors, including work that showcased the use of National Water Model (NWM) Predictions, spatiotemporal methodologies and analytics for the study of extreme weather, the use of machine learning with GIS, and the visual analytics of big data.

Four panels on Friday discussed sensors and infrastructure, analysis and interpretation, applications and case studies for cities, health, and environment, and future directions and implications for civic engagement. The 20 panelists presented highlights of their research and addressed questions from the audience, including questions about privacy, openness and interoperability, uncertainty and measurement accuracy, and predictive biases.

David DiBiase of Esri and Matthew Wilson of the University of Kentucky offered their closing remarks on the convergence of data science and GIScience, noting the urgency for more collaboration between geographers, GIScientists, and data scientists. Two Harvard students were awarded the Fisher Prize for their posters: Jennifer Horowitz in the undergraduate category and Yousef Awaad Hussein in the graduate student category.

by Wendy Guan, Jason Ur and Matt Wilson