Using Floating Car Data (FCD) to Infer Mobility Patterns

With the widespread use of location sensing technologies such as GPS-enabled vehicles, huge volumes of vehicle trajectory data are increasingly generated. The growing availability of such data opens up new opportunities for performing more sophisticated and comprehensive spatial and temporal analyses for planning and management of transportation systems. One of the most useful types of analysis in this context is traffic data clustering, which can help in understanding and revealing valuable insights into urban mobility patterns and travel behavior. In this thesis, a six-day dataset of floating car data (FCD) from Munich city is used to extract meaningful urban mobility patterns. Hierarchical clustering analysis is used first to spatially cluster the trips in each day based on the coordinates of their origin and destination points, such that each cluster contains the trips that travel from one specific origin zone to another destination zone. Next, an innovative tool, called Relative Deviation Area (RDA), is introduced to help in understanding travel behavior in the resulting clusters. RDA aims to find the relative area by which a given trajectory is deviating from a referential trajectory (typically the least-cost path). RDA is computed for each trip in each cluster on each day. This is followed by investigating the relationship between trip average speed (V) and RDA for each day using Kernel regression method. The resulting regression curves are found sensible and consistent throughout all days, which indicates a potential association between the two variables. In addition, the relationship is temporally investigated at peak and off-peak periods. V values at peak periods are found to be lower than those at off-peak periods for the same value of RDA. Another case is tested where only private cars are considered, excluding all other vehicle types like taxicabs and trucks. The results showed that RDA values in private cars case are higher than those in all vehicle types case. The output illustrates the potential of using Big Data to infer mobility patterns and travel behavior. The developed RDA tool is expected to have several applications in different fields such as urban and transportation planning, transportation demand management, and traffic monitoring.     «

With the widespread use of location sensing technologies such as GPS-enabled vehicles, huge volumes of vehicle trajectory data are increasingly generated. The growing availability of such data opens up new opportunities for performing more sophisticated and comprehensive spatial and temporal analyses for planning and management of transportation systems. One of the most useful types of analysis in this context is traffic data clustering, which can help in understanding and revealing valuable ins...     »