Applying a deep learning approach for transportation safety planning by using high-resolution transportation and land use data

摘要: Abstract Analytical Transportation Safety Planning (TSP) is an important concept for integrating and improving both planning safety achieving better policies decision making. In the recent decade, considerable efforts have been devoted to providing prediction results with consideration of zonal systems, mathematical methods, input variables, etc. previous studies, transportation land use data widely used as predict crashes. Meanwhile, studies required all variables be aggregated at level. With aggregation process, collected fell into low resolution lost details, which may introduce accuracy even biases. The primary objective this study validate viability applying a deep learning approach crashes TSP high-resolution data. A framework collecting first introduced. Then, architecture convolutional neural network (CNN) adopted traffic To proposed method, empirical conducted method compared three counterparts: two statistical models (i.e., negative binomial model spatial Poisson lognormal model) traditional machine artificial network) using low-resolution that are based on zones). indicate could provide significantly higher than conventional data, validates detailed crash prediction. It expected in new valuable insights future directions planning.