Moudon, Anne VernezKang, Min Gyu2019-08-142019-08-142019Kang_washington_0250E_20298.pdfhttp://hdl.handle.net/1773/44461Thesis (Ph.D.)--University of Washington, 2019The goals of active and safe transportation can be achieved by creating a pedestrian-friendly environment. Since walking trips are more likely to be observed in dense urban areas where motorized travel is congested, a safe environment from motorized vehicles is crucial to protecting pedestrians and promoting walking. Thus, identifying locations where pedestrians are most vulnerable is important to further promoting this environmentally friendly and healthy mode of travel. The characteristics of the built environment at these locations help capture attributes that can affect the risk of crashes: for example, development densities and some related land uses attract pedestrian travel, while sidewalks and traffic signals can protect them from colliding with vehicles. As a result, quantifying built environment attributes around a crash-risk location is an important component in modeling pedestrian crashes. A variety of data and methods have been used to identify crash-risk locations in different studies, which have limited comparability across studies and caused complications in interpreting the results. To date, most studies measured the overall characteristics of environments around potential crash locations for pedestrian crash modeling. However, measuring the built environment along an actual pedestrian route can more precisely capture the characteristics related to the risk of a crash than those derived from location-based approaches. Objectively measured mobility data coming from such devices as global positioning system (GPS) and accelerometry have the potential to overcome limitations in location-based analyses. Processing the massive GPS and accelerometer datasets to reconstruct mobility patterns in terms of trips and travel modes requires robust computational power and sophisticated algorithms. Few studies have focused on understanding the details on how these algorithms process data for the purposes of quantifying travel behaviors. In this dissertation, analyses first used a location-based pedestrian safety approach that combined built environment and crash data to identify crash-risk locations and to model pedestrian-motor-vehicle collisions. More specifically, a new protocol was developed to provide a useful tool for identifying unique pedestrian crash-risk locations at intersection and non-intersection areas. Second, the factors affecting pedestrian crashes were evaluated using fine-grained built environment data. Lastly, the automated travel behavior detection algorithm PALMS (Personal Activity Location Measurement System) was assessed: PALMS approach to translate objective measures of individual mobility patterns (e.g., GPS and accelerometry) into trips and travel modes was compared to trips recorded in travel diaries. Studies in this dissertation contribute to the creation of consistent spatial analysis units for location-based pedestrian crash models, which makes it possible for empirical results to be comparable. A cost-effective method is offered to identify unique crash-risk locations. The dissertation also contributes to the literature by showing that factors affecting pedestrian crashes at intersection locations differed from those of non-intersection locations. It provides advanced visualization approaches to interpret empirical model results which can be used to prioritize safety countermeasures according to the characteristics of potential crash locations. Also investigated is the potential of device-collected mobility data from GPS and accelerometers to help identify individual travel modes, and to detect travel routes that can be used for quantifying the built environment attributes. The PALMS algorithm was found to better classify vehicular than pedestrian travel. Lastly, the methods developed to assess PALMS can be generalized and can serve to evaluate different approaches to travel mode classification.application/pdfen-USnoneActive TransportationAlgorithmGISGPSLand UsePedestrian SafetyUrban planningTransportationLand use planningUrban planningThesis