Irregular Travel Groups Detection Based on Cascade Clustering in Urban Subway

Travel smart cards record passengers’ travel histories, which makes it possible to study personal traveling behaviors and passengers’ mobility patterns. The existing researches on smart card data pay less attention to those who beg, steal or busk during traveling, and they are called irregular passengers in this paper. Moreover, the group consisted of irregular passengers is called irregular travel group. In this paper, we propose an approach to recognize irregular travel groups based on cascade clustering. Firstly, passengers’ travel sequences of active state in hours are extracted as representations of travel patterns according to the records of smart cards in a continuous time period. The sequences are clustered with the K-means algorithm to detect irregular passengers. Then, travel similarities between the derived irregular passengers are measured and irregular travel groups are recognized based on DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm. The proposed method is applied to real Beijing subway smart card data, and the results are validated with the data derived from SINA Micro-blogs. Experimental results show that the detected irregular passengers’ movement areas are consistent with the ground truth.