Speaker: Daniel Work, University of Illinois at Urbana-Champaign
Abstract: This talk considers the emerging problem of quantifying the resilience of the city-scale road network to extreme events. Compared to existing traffic performance reports that operate on monthly or annual data, a framework is proposed to quantify the transportation network state in terms of the hourly average traffic pace on individual road segments across an urban region. The high dimensionality of the traffic state is reduced and outlier hours are identified using a sparse matrix decomposition approach. The method is implemented in New York City over a four-year period from 2010-2013 using traffic data inferred from 697 million taxi trips, which is available for download at http://publish.illinois.edu/dbwork/open-data/. The method allows new traffic dynamics in the aftermath of disasters to be discovered. For example, it illustrates that the evacuation process prior to Hurricane Sandy caused only minor disruptions, but significant delays were encountered during the post-disaster reentry process.