Authors: Bandana Kar*, Oak Ridge National Laboratory
Topics: Geographic Information Science and Systems, Hazards, Risks, and Disasters
Keywords: Mobility pattern, trajectory mining, emergency management
Session Type: Paper
Presentation File: No File Uploaded
During natural hazards, many critical infrastructures, including transportation networks, are crippled which subsequently makes undertaking emergency management efforts such as search and rescue efforts, evacuation routing, and relief efforts difficult. With availability of heterogenous big data (e.g., texts, images and videos from social media sites, cell phone records, Global Positioning System (GPS) data), data-intensive machine learning approaches have been successful in predicting human mobility pattern. Using road network data from OpenStreetMaps and taxi trajectory data from the New York City Taxi and Limousine Commission for New York City for the month of October and November of 2012, in this study mobility patterns were determined using trajectory data mining approach before and after hurricane Sandy (2012). The purpose was to (i) explore the mobility patterns before and after the hurricane within the City and (ii) identify potential routes following the hurricane based on existing mobility pattern to help with emergency management activities. Next step of this study involves implementation of reinforcement learning approaches to forecast potential patterns and evacuation routes.