Analysis of Traffic Crashes Involving Pedestrians Using Big Data: Investigation of Contributing Factors and Identification of Hotspots

https://doi.org/10.1111/risa.127852017PDFJournal ArticleCostsData analysisHigh risk locationsMathematical modelsPedestrian-vehicle crashesRisk assessmentSpeed dataTraffic dataUnited StatesConnected Cities for Smart Mobility toward Accessible and Resilient Transportation Center (C2SMART)Xie, KunOzbay, KaanKurkcu, AbdullahHong, YangConnected Cities for Smart Mobility toward Accessible and Resilient Transportation Center (C2SMART)US Transportation CollectionThis article reports on a study undertaken to explore the use of big data in pedestrian risk analysis, specifically strategies to investigate contributing factors and hotspots (areas with high incidence of crashes). The authors analyzed large amounts of data of Manhattan (New York) from a variety of sources, including taxi trips, subway turnstile counts, traffic volumes, road network, land use, sociodemographic, and social media data. The authors then created a system of grid cells as the basic geographical units of analysis. The cost of each crash, weighted by injury severity, was assigned to the grid cells based on the relative distance to the crash site using a kernel density function. The authors developed a tobit model to relate grid-cell-specific contributing factors to crash costs that are left-censored at zero. The authors calculated the potential for safety improvement (PSI) by using the actual crash cost minus the cost of \u201csimilar\u201d sites; this was used as a measure to identify and rank pedestrian crash hotspots. The authors contend that their hotspot identification method takes into account two factors that are generally ignored, i.e., injury severity and effects of exposure indicators. They conclude with a brief discussion of the possibilities of using big data to enable more precise estimation of the effects of risk factors and to support large-scale hotspot identification. By using a cell-structured framework, researchers can incorporate richer and more diversified data sets into safety modeling.82