Insight into traffic security: A correlation discovery of urban spatial features and traffic flow patterns

With the rapid development of urbanization, urban traffic security problems become increasingly prominent, and traffic accidents occur frequently. It is becoming increasingly critical to use data mining methods to solve actual problems in traffic security. But we don’t just rely on the analysis of urban traffic surface operation rules, it is to study the mechanism of traffic operation. Therefore, in this paper, we aim to discover the correlation between spatial features and traffic flow patterns in urban regions to improve traffic security. In order to obtain deeper spatial semantic meanings for every urban region, we propose a spatial feature method based on regional hierarchy, which fuses the hierarchical structure of region and POI (point of interest) information. For traffic flow patterns, we propose a self-representation learning optimization method to find the similarity of region traffic flow. Then, we match the spatial features and flow patterns to discover the correlation of them. Experiments show that our approaches are effective.