Seminar Series: Dr. Vanessa Frias-Martinez Will Give a Talk on Open and Crowdsourced Data for the Prediction of Perceived Cycling Safety at City Scale

Dr. Vanessa Frias-Martinez has been invited to participate in the Department Seminar Series by Dr. Leila DeFloriani. She will give her seminar on October 3 at 4pm in 1158 LeFrak Hall. Please arrive early to meet her and get settled. 

Title

Open and Crowdsourced Data for the Prediction of Perceived Cycling Safety at City Scale

Abstract

The benefits of cycling are well studied: reduction in urban pollution; savings in healthcare costs due to daily exercise; lower stress levels and obesity rates; and improvements in workforce accessibility for low-income communities, among others. As a result, cities have created bike lanes, promoted bike-share systems and supported bike to work programs in the past years. However, building bicycle-friendly communities and increasing ridership in cities will require an accurate understanding of how urban cycling safety is perceived by residents at both street- and trip-scale. Evaluating perceived cycling safety at the street segment level will allow to identify safety barriers and specific street improvements; while the analysis of cycling safety at the trip level will inform of spatial accessibility challenges that might prevent residents from reaching places due to cycling safety concerns. Current approaches to understanding perceived cycling safety usually require access to expensive sensors that measure variables such as daily traffic or average speed, thus highly limiting the understanding of cycling safety to cities with economic resources, and to only a handful of streets where such sensors are deployed. In this talk, I will present a new approach that exclusively uses open and crowdsourced data to make cycling safety information accessible to a larger number of cities and for a larger number of streets. First, I will describe an open-source, crowdsourced rating platform that we have developed to help cities gather safety perceptions from cyclists. Second, I will present novel methods to predict perceived cycling safety at the street segment level using open and crowdsourced data; and discuss the role that social and built environment features play in cycling safety perceptions. Third, I will describe LSTM-based approaches that incorporate spatio-temporal dependencies to aid in the prediction of perceived cycling safety at the trip level. We expect that our approach will lower the bar in the access to comprehensive and accurate cycling safety information for many cities worldwide.

Bio

Dr. Vanessa Frias-Martinez is an assistant professor in the iSchool and UMIACS, and an affiliate assistant professor in the Department of Computer Science at the University of Maryland, College Park. She also leads the Urban Computing Lab. Frias-Martinez's research areas are data-driven behavioral modeling and spatio-temporal data mining. Her research focuses on the use of large-scale ubiquitous data to model the interplay between human mobility and the physical environment. Specifically, Frias-Martinez develops methods to model and predict human behaviors in different contexts and creates tools to aid decision makers in areas such as disaster preparedness and response or urban planning. Her research is funded by the National Science Foundation, including a CAREER Award, and by the World Bank.  Before coming to UMD, she spent five years at Telefonica Research, and completed internships at Google and the United Nations Development Program (UNDP). Frias-Martinez received her PhD in Computer Science from Columbia University in 2008.